Neuralink: FDA compliance and adverse event reporting in PRIME human trial 2026

PRIME Study Status 2026: Enrollment Metrics vs

As of January 28, 2026, Neuralink Corp. has enrolled 21 human subjects in its clinical trials worldwide, a figure that represents a significant acceleration in surgical velocity following a stagnant 2024. yet, an analysis of federal filings and public statements reveals a persistent gap between the company’s aggressive projection and verified clinical reality. While the PRIME Study (Precise Robotically Implanted Brain-Computer Interface) successfully expanded beyond its initial US cohort into the United Kingdom and Canada by late 2025, the that early mechanical failures, specifically thread retraction problem in the patient, forced a six-month enrollment freeze that permanently offset Elon Musk’s initial deployment timeline.

The current enrollment count of 21 participants, confirmed by Fidelity reports and Neuralink’s January 2026 updates, marks a sharp increase from the 12 patients reported in September 2025. This surge suggests that the FDA and international regulatory bodies like the UK’s MHRA cleared Neuralink for a rapid “catch-up” phase in Q4 2025. Yet, this operational ramp-up occurs against a backdrop of missed milestones. Musk’s stated goal for 2024 was to implant “high single digits” (approximately 8, 10 patients); the actual 2024 close-out number was two. The 2025 target, revised to “20 to 30” new implants, was met only at the lowest margin of the threshold, with the majority of surgeries compressed into the final quarter of the year.

Enrollment Velocity and Target Discrepancies

The trajectory of the PRIME study (registered under identifiers including NCT06429735) demonstrates a non-linear growth pattern characteristic of trials plagued by early safety signals. The initial protocol anticipated a steady cadence of surgeries to assess the N1 implant’s safety. Instead, the timeline shows a “halt-and-sprint” cadence.

The primary disruption occurred immediately following the January 2024 implantation of Patient 1, Noland Arbaugh. By February 2024, Neuralink engineers detected that 85% of the thread electrodes had retracted from the motor cortex, severely degrading the signal-to-noise ratio. This adverse event, while not presenting a direct safety threat to the patient’s life, undermined the device’s efficacy claims and necessitated a software-based mitigation strategy. Consequently, the second patient, “Alex,” was not implanted until July 2024, a five-month gap that derailed the 2024 enrollment.

Table 1: Neuralink PRIME Study Enrollment: vs. Verified Actuals (2024, 2026)

Period	Stated Target (Musk/Neuralink)	Verified Actual Enrollment	Cumulative Total	Status
Q1 2024	1 Patient	1 (Noland Arbaugh)	1	On Target
Q2, Q3 2024	3, 4 Patients	1 (Alex)	2	Missed (Paused for thread retraction)
Q4 2024	10 Total (Cumulative)	0 New (Total 2)	2	Failed (80% deficit)
Q1, Q2 2025	Start of “20, 30” New	3 New (RJ + 2 others)	5	Delayed
Q3 2025	Accelerated Ramp	7 New	12	Accelerating (Expansion to Canada)
Q4 2025	Complete 2025 Cohort	9 New	21	Met Low Target (Surge in UK/US)

The “Catch-Up” Surge of Late 2025

The data shows a distinct shift in operational tempo beginning in September 2025. Following the clearance of the thread retraction mitigation, which involved implanting threads deeper into the cortex to prevent migration, Neuralink quadrupled its patient count in under six months. By September 10, 2025, the company confirmed 12 active patients. By January 28, 2026, that number hit 21. This implies a surgical cadence of roughly two implants per month throughout late 2025, a rate significantly higher than the single surgery performed in the half of 2024.

This acceleration was fueled by the opening of international sites. The “GB-PRIME” study in the United Kingdom, launched in July 2025 at University College London Hospitals, contributed at least seven patients to the global total by early 2026. Similarly, the “UAE-PRIME” feasibility study (NCT06992596) and Canadian sites began active recruitment, allowing Neuralink to bypass the bottleneck of US-only enrollment. This fan-out strategy insulated the company from single-regulator slowdowns; when US enrollment stabilized, UK and UAE sites absorbed the volume required to meet investor milestones.

“The reason we do clinical trials and early feasibility trials is to uncover these sort of problem as early as possible before they get marketed. We rolled up our sleeves and found various different ways for Noland to recover his performance.”
, DJ Seo, Neuralink Co-founder (May 2024)

Regulatory of the 2024 Stumble

The gap between the 2024 target (10 patients) and the actual result (2 patients) is not a logistical failure a regulatory artifact. The FDA’s investigational device exemption (IDE) allows for iterative expansion based on safety data. The thread retraction problem in Patient 1 triggered a “safety pause” method. Although not a formal clinical hold, the need to demonstrate a surgical or software fix to the FDA before proceeding to Patient 2 created the five-month void in the 2024 timeline.

Investigative analysis of the 2025 cohort suggests that the FDA accepted Neuralink’s mitigation strategy, implanting threads at a depth of 3-4mm rather than the original 2-3mm, as sufficient to resume enrollment. yet, the long-term efficacy of this fix remains under scrutiny. While Patient 2 (Alex) did not report immediate retraction problem, the sample size of 21 patients as of 2026 is large enough to determine if the “thread pull-back” is a widespread material failure or a procedure-specific anomaly. The rapid enrollment of 19 patients in 2025 indicates that Neuralink has likely moved to a “confirmatory” phase of the feasibility study, where the focus shifts from basic safety to establishing a consistent surgical protocol for the R1 robot.

Demographic and Site Distribution

The expansion of the PRIME study has also diversified the patient demographic. Initially restricted to US residents with cervical spinal cord injuries (SCI) or ALS, the 2025 cohort includes international participants. Verified reports indicate:

• United States: Approximately 10, 12 patients. Primary site: Barrow Neurological Institute (Phoenix, AZ).
• United Kingdom: 7 confirmed patients (GB-PRIME). Primary site: University College London Hospitals.
• Canada: 2 confirmed patients. Site: University Health Network (Toronto).
• UAE: Active recruitment for UAE-PRIME, aiming for 10 participants.

This geographic distribution serves a dual purpose: it validates the R1 robot’s transportability and functionality in diverse hospital environments, and it creates a redundant regulatory framework. If the FDA tightens restrictions due to a new adverse event, the MHRA (UK) or Canadian authorities may allow the trial to continue, ensuring the data pipeline remains open. The 2026 enrollment metrics, therefore, reflect a company that has successfully engineered its way out of a single point of regulatory failure, even if it has yet to prove the long-term durability of its device in the expanded cohort.

The 85% Signal Collapse: Anatomy of the N1 Failure

In May 2024, Neuralink Corp. confirmed a significant mechanical failure in its human clinical trial participant, Noland Arbaugh. Approximately 85% of the electrode-bearing threads implanted in Arbaugh’s motor cortex retracted from the brain tissue, severing the neural connection required for the device’s primary function. The N1 implant, designed with 64 flexible polymer threads containing a total of 1, 024 electrodes, relies on physical proximity to neurons to detect action chance. The retraction event left only roughly 15% of the electrodes functional, resulting in a precipitous drop in the system’s bits-per-second (BPS) data transmission rate, the core metric for cursor control accuracy and speed.

The failure occurred within weeks of the January 2024 surgery. While the device remained electrically safe and powered, the physical decoupling of the threads rendered the majority of the channels silent. Internal data reviewed by the FDA indicated that the threads, which are thinner than a human hair and inserted to a depth of 3 to 5 millimeters, were dislodged as the brain moved within the intracranial space. This phenomenon, known as micromotion, is a known challenge in brain-computer interface (BCI) engineering, yet the magnitude of the retraction in the PRIME study’s subject exceeded pre-clinical projections derived from animal models.

Root Cause: Pneumocephalus and Intracranial Shift

Post-incident analysis identified pneumocephalus, the presence of trapped air inside the cranial cavity, as a primary contributing factor to the thread retraction. During the surgical procedure, the craniectomy (removal of a portion of the skull) created a void that allowed air to enter. Following the closure of the surgical site, this air pocket permitted the brain to shift significantly within the skull, moving up to three times more than Neuralink engineers had modeled based on porcine and non-human primate data.

The N1 threads are not anchored to the skull or the dura mater; they float freely in the brain tissue to minimize damage during pulsation. yet, the excessive range of motion caused by the pneumocephalus pulled the threads out of the cortex. The retraction was not a uniform withdrawal a chaotic decoupling, with threads pulling out completely while others remained partially inserted functionally degraded. The mechanical mismatch between the rigid implant casing (sited in the skull) and the soft, shifting brain tissue under conditions of high intracranial mobility proved catastrophic for the initial thread depth.

Table 2. 1: N1 Implant Thread Status, Patient 1 (Noland Arbaugh)

Metric	Pre-Incident Status (Jan 2024)	Post-Retraction Status (May 2024)	Impact
Total Threads	64	~9-10 (Estimated )	85% Loss of Hardware Inputs
Active Electrodes	1, 024	~150	Severe Signal Resolution Drop
Insertion Depth	3-5 mm	Retracted/Displaced	Loss of Neural Proximity
Cursor Control (BPS)	Record High	Significant Decline	Algorithm Recalibration Required

Software Remediation and Algorithm Sensitivity

Faced with a hardware failure that would require explantation or revision surgery, Neuralink opted for a software-based remediation strategy. The company’s engineering team rewrote the recording algorithm to increase sensitivity to neural population signals. Originally, the system was calibrated to detect spikes from specific individual neurons (single-unit activity). With the loss of proximity, the signal-to-noise ratio degraded, making single-unit isolation impossible on the retracted channels.

The updated decoding model shifted focus to “multi-unit activity” and local field chance, broader, aggregate electrical signals generated by groups of neurons. By amplifying these weaker signals and re-mapping the vector decoding matrix, engineers were able to restore cursor control. Although the resolution of the raw neural data was permanently compromised, the functional output (cursor movement) recovered to levels method, and eventually superseding, Arbaugh’s initial performance. This pivot demonstrated the plasticity of the decoding software highlighted a serious reliance on algorithmic compensation for mechanical instability.

Surgical Protocol Overhaul for Patient 2

The retraction incident necessitated immediate revisions to the surgical protocol for the second participant, Alex, implanted in July 2024. To mitigate the risk of pneumocephalus-induced retraction, Neuralink implemented three specific changes:

1. Skull Sculpting: The craniectomy technique was refined to minimize the gap between the implant and the brain surface, reducing the volume of air that could be trapped.
2. CO2 Management: Enhanced measures were introduced to maintain normal carbon dioxide levels in the blood during surgery, which helps regulate intracranial pressure and brain volume.
3. Increased Depth: The insertion depth for the threads was increased from the original 3-5 mm range to 8 mm. This deeper anchoring places the electrodes further into the cortical, providing a greater mechanical buffer against brain shift.

Data from the second patient, released in August 2024, showed zero thread retraction one month post-surgery, validating the efficacy of the revised surgical parameters. The absence of retraction in the second subject suggests that the failure in Patient 1 was largely procedural rather than a fundamental flaw in the thread flexibility itself, provided that insertion depth and air gap management are strictly controlled.

Regulatory Oversight and Pre-Clinical Signals

Investigative scrutiny has revealed that the FDA and Neuralink were aware of the chance for thread retraction prior to the human trials. Adverse event reports from animal testing, specifically in pigs and monkeys, contained instances of thread migration and granuloma formation. Reuters reported in May 2024 that Neuralink had deemed the risk “low” enough to proceed without a design overhaul, a decision that the FDA accepted during the approval process for the PRIME study.

The decision to proceed with the 3-5 mm depth in the patient, even with the animal data, reflects a calculated risk in the translation from animal models to human anatomy. The human brain is significantly larger and subject to different gravitational and inertial forces than the brains of macaques or pigs. The 85% failure rate in the human subject show the limitations of animal proxies for predicting mechanical device-tissue interactions in the human cortex. The FDA has since intensified its monitoring of the PRIME study, requiring detailed reporting on thread stability for all subsequent participants.

Surgical Mitigation Strategies: Efficacy of 8mm Insertion Depth in Patient 2

Following the significant mechanical failure observed in the PRIME study participant, Neuralink Corp. implemented a revised surgical protocol for its second human subject, identified as “Alex.” The primary objective of these modifications was to counteract the “thread retraction” phenomenon that rendered 85% of the electrodes in the patient, Noland Arbaugh, ineffective within weeks of implantation. In July 2024, surgeons at the Barrow Neurological Institute executed this updated protocol, which centered on increasing the insertion depth of the N1 implant’s polymer threads and minimizing the physical gap between the device and the brain surface.

The 8mm Depth Protocol

The most serious deviation from the initial surgical plan was the adjustment of electrode insertion depth. In Patient 1, threads were implanted to a depth of 3mm to 5mm, targeting the motor cortex’s gray matter. Post-surgical analysis revealed that this depth was insufficient to anchor the threads against the brain’s natural pulsatile motion and the air-gap-induced shifting (pneumocephalus) that occurred post-operation. For Patient 2, Neuralink obtained FDA approval to increase the insertion depth to 8mm.

This 60% to 160% increase in depth was designed to anchor the threads more securely within the neural tissue, mitigating the risk of the threads “backing out” as the brain moved. also, the surgical team employed “skull sculpting,” a technique involving the intentional contouring of the skull surface to allow the implant casing to sit flush against the dura mater. This reduction in the gap between the device and the brain was calculated to eliminate the tension on the threads caused by the fluid previously observed in Patient 1.

Comparative Surgical Parameters

The following table outlines the key operational differences between the and second PRIME study implantations, highlighting the escalation in risk mitigation measures.

Parameter	Patient 1 (Noland Arbaugh)	Patient 2 (Alex)	Delta / Rationale
Surgery Date	January 2024	July 2024	+6 Months (Root Cause Analysis)
Thread Insertion Depth	3mm , 5mm	8mm	Increased anchor stability
Skull Preparation	Standard Craniotomy	Contoured “Skull Sculpting”	Minimize implant-to-brain gap
Intraoperative CO2 Control	Standard	Strict Normalization	Reduce brain volume fluctuation
Retraction Outcome	85% of threads retracted	0% retraction (at 4 months)	Full stabilization achieved

FDA Oversight and Approval

The transition to an 8mm insertion depth required an amendment to Neuralink’s Investigational Device Exemption (IDE). In May 2024, the FDA granted approval for the second patient’s surgery, validating the company’s root cause analysis which attributed the initial failure to mechanical instability rather than biological rejection. The agency’s review focused on the safety of deeper cortical penetration, specifically the risk of engaging white matter tracts or inducing vascular damage. The approval indicated that the regulatory body accepted the risk-benefit profile of the deeper insertion to ensure device efficacy.

Outcome Analysis: Zero Retraction

Data collected through late 2025 confirms the efficacy of the revised surgical strategy. Unlike the rapid signal degradation seen in Patient 1, Patient 2 exhibited no evidence of thread retraction. Neuralink’s August 2024 update reported that the threads in Alex’s motor cortex remained stable, maintaining a consistent number of electrodes. This stability allowed for immediate and sustained high-performance cursor control, with Alex successfully using the interface to design 3D objects in Fusion 360 and play complex -person shooter games like Counter-Strike 2.

“Promisingly, we have observed no thread retraction in our second participant… The threads have stabilized, and the performance has remained strong.” , Neuralink PRIME Study Update, August 2024.

The absence of retraction in Patient 2 serves as a validation of the mechanical anchoring hypothesis. By reducing the “slack” in the system through skull sculpting and increasing the friction/anchoring force via deeper insertion, the surgical team successfully decoupled the implant’s motion from the brain’s natural movements. This result suggests that the 85% failure rate in the trial was a preventable mechanical error rather than an inherent flaw in the flexible thread technology itself.

FDA Form 483 Inspection Findings: Quality Control Lapses at Fremont Animal Facility

Between June 12 and June 22, 2023, the U. S. Food and Drug Administration (FDA) conducted a detailed inspection of Neuralink’s animal research facility in Fremont, California. This inspection, which occurred less than a month after the FDA granted approval for the PRIME human clinical trial, resulted in the issuance of a Form 483. The document “objectionable conditions” and significant lapses in quality control that directly impact the integrity of the preclinical data used to validate the N1 implant’s safety.

Inspection Timeline and Scope

The FDA’s inspection focused on the Fremont facility’s adherence to Good Laboratory Practice (GLP) regulations, which are federal mandates designed to ensure the quality and integrity of non-clinical safety data. Unlike the company’s Texas facility, which inspectors found to be compliant, the California site failed to produce serious maintenance and calibration records for equipment used in pivotal animal studies.

Inspection Parameter	Details
Inspection Dates	June 12 , June 22, 2023
Facility Location	Fremont, California (Animal Research)
Regulatory Action	Form 483 Issued (Objectionable Conditions)
Key Deficiencies	Missing calibration records, unsigned QA reports, undocumented protocol deviations

Specific Quality Control Violations

The Form 483 detailed specific instances where Neuralink failed to maintain the rigorous standards required for medical device testing. Investigators discovered that the facility absence calibration records for a pH meter used in one of the studies. More serious, for another study, seven separate instruments, including a important signs monitor, had no record of being calibrated. These instruments are essential for monitoring the physiological stability of test subjects during complex neurosurgical procedures.

Beyond equipment failures, the inspection revealed widespread problem in data reporting. Quality assurance officials at Neuralink had not signed off on the final study report for the inspected experiments. also, the facility failed to document deviations from approved. In a regulated environment, any from the pre-approved study plan must be meticulously recorded to assess its impact on the data’s validity. The absence of these records creates a “black box” around the animal data, making it difficult to verify whether adverse events in test subjects were accurately captured or if they were artifacts of uncalibrated equipment.

Regulatory and Data Integrity

The timing of these findings is significant. The inspection took place shortly after the FDA’s May 2023 approval of the PRIME human trial. While the FDA stated in December 2024 that the violations did not “undermine the device’s safety” enough to warrant a stop to the human trials, the lapses raise questions about the baseline safety data derived from animal models. The Physicians Committee for Responsible Medicine (PCRM) had previously filed complaints alleging that Neuralink’s animal testing practices were rushed and sloppy, leading to unnecessary suffering and compromised data. The FDA’s findings of “objectionable conditions” corroborate concerns that the pressure to meet aggressive timelines may have eroded adherence to standard operating procedures.

“These problem show a absence of attention to detail… This certainly is a signal that the company needs to be vigilant about certain practices.”
, Jerry Chapman, Senior Quality Expert, Redica Systems (Feb 2024)

The FDA opted for “voluntary remediation” rather than issuing a Warning Letter or an Official Action Indicated (OAI) classification. This regulatory stance allows Neuralink to correct the problem without halting its ongoing operations. yet, the absence of a verified calibration history for important signs monitors means that historical data on animal physiological responses during implantation cannot be fully audited for accuracy. If a test subject experienced a spike in blood pressure or heart rate during surgery, and the monitor was uncalibrated, the true magnitude of that event remains unknown.

20-Point Fan-Out: Inspection Analysis

To fully contextualize the impact of the Form 483, we examined 20 serious aspects of the inspection and its aftermath:

1. Inspection Window: The FDA was on-site for 10 days in June 2023.
2. Primary Citation: Failure to maintain calibration records for serious instruments.
3. Specific Device 1: A pH meter used in bio-compatibility studies absence records.
4. Specific Device 2: A important signs monitor was among seven uncalibrated instruments in a single study.
5. QA Failure: Final study reports were not signed by the Quality Assurance unit.
6. Protocol Deviations: Changes to approved experimental procedures were not documented.
7. Facility Comparison: The Texas facility was inspected and found compliant; Fremont was not.
8. Human Trial Impact: The inspection occurred post-approval; the FDA did not rescind the IDE (Investigational Device Exemption).
9. PCRM Role: Prior complaints by PCRM regarding “hack jobs” and rushed surgeries triggered scrutiny.
10. Adverse Events: Uncalibrated monitors complicate the retrospective analysis of animal adverse events.
11. Regulatory Status 2024: As of Dec 2024, the FDA considers the problem “voluntary remediation” items.
12. Data Validity: The absence of QA sign-off technically renders the final reports “draft” quality under strict GLP interpretation.
13. Surgical Robot: The inspection focused on the biological side; robot calibration was not the primary citation in this 483.
14. Animal Subjects: The studies involved rhesus macaques and pigs.
15. Response: Neuralink did not publicly release a detailed rebuttal complied with the voluntary remediation request.
16. Investigator: The FDA sent specialized investigators to the site, indicating a high-level audit.
17. Documentation: The core failure was “record keeping” rather than proven device failure.
18. Safety Baseline: The unverified animal data serves as the safety baseline for the current human subjects.
19. widespread problem: The presence of multiple uncalibrated instruments suggests a widespread lapse in the equipment management program.
20. Future Oversight: The FDA has signaled continued monitoring of the PRIME trial’s safety reports of these findings.

The Fremont facility’s inability to produce basic calibration records for important signs monitors presents a serious gap in the data trail. In high- neurological research, the difference between a successful implantation and a catastrophic physiological reaction is frequently measured in subtle changes in important signs. Without verified calibration, the “safety” signal from these animal studies relies on the assumption that the equipment was working correctly, rather than the proof that it was.

SECTION 5 of 22: Hazardous Material Compliance: DOT Fines for Xylene and Biological Waste Transport

Federal Pipeline and Hazardous Materials Safety Administration (PHMSA) Investigation

In January 2024, the U. S. Department of Transportation (DOT) concluded a formal investigation into Neuralink Corp., resulting in a civil penalty for violations of federal hazardous material transport laws. The probe, executed by the Pipeline and Hazardous Materials Safety Administration (PHMSA), confirmed that the company had failed to adhere to serious safety regarding the movement of hazardous chemicals and waste. Records finalized in early 2024 indicate that Neuralink agreed to pay a total fine of **$2, 480**. While the monetary value appears nominal relative to the company’s multi-billion dollar valuation, the infractions reveal a specific breakdown in logistics compliance during the serious period preceding the PRIME human trials. The investigation was triggered following a complaint filed by the Physicians Committee for Responsible Medicine (PCRM), which provided the DOT with internal emails suggesting a pattern of negligence in handling explanted neural devices.

Confirmed Violations: Xylene and Hazmat Registration

The PHMSA inspection, conducted in February 2023 at Neuralink’s facilities in Texas and California, substantiated two primary violations of the Hazardous Materials Regulations (HMR)., investigators found that Neuralink was transporting **Xylene**, a flammable, colorless liquid solvent used in tissue processing and histology, without proper packaging. Xylene is classified as a hazardous material due to its flammability and chance to cause serious health effects, including central nervous system depression, respiratory irritation, and, in high concentrations, death. The DOT citation noted that the packaging used for this volatile chemical was insufficient to ensure safety during transit, violating 49 CFR packaging requirements. Second, the agency determined that Neuralink had **failed to register as a transporter of hazardous materials**. Under federal law, any entity that ships or transports specific quantities or types of hazardous materials must register with the DOT and pay a fee to fund emergency response planning grants. Neuralink’s failure to register indicated a lapse in administrative compliance, operating outside the federal oversight framework designed to track the movement of dangerous goods.

Biological Waste and Pathogen Transport Allegations

The DOT investigation was originally prompted by more severe allegations regarding the transport of biological waste. PCRM’s complaint internal emails from 2019 involving the University of California, Davis (UC Davis), where Neuralink previously conducted primate research. The correspondence described incidents where Neuralink employees allegedly transported explanted neural devices, removed from the brains of rhesus macaques, in open, unsealed containers. University staff expressed urgent concern in these emails, noting that the hardware had not been sterilized and could chance carry **Herpes B virus** or antibiotic-resistant pathogens like *Staphylococcus* and *Klebsiella*. One email from a UC Davis employee stated, “Since the hardware components of the explanted neural device are not sealed and it was not disinfected prior to leaving the university, this presents a hazard for anyone chance coming in contact with the device.”

DOT Findings on Biological Agents

While the PHMSA inspection confirmed the Xylene and registration violations, the agency’s final report noted that investigators **did not find evidence** during the 2023 inspection that Neuralink was currently shipping infectious substances in violation of regulations. The confirmed fines were strictly attached to the chemical solvent (Xylene) and the registration failure, rather than the transport of Category B infectious substances. yet, the investigation highlighted the operational friction between the company’s rapid development ethos and the rigid safety required for handling biological risks.

Summary of DOT/PHMSA Enforcement Action (2023-2024)

Violation Type	Specific Material	Regulatory Finding	Status
Improper Packaging	Xylene (Flammable Liquid)	Packaging failed to meet 49 CFR standards for hazardous solvents.	Confirmed / Fined
Registration Failure	General Hazardous Materials	Company failed to register as a hazmat transporter with DOT.	Confirmed / Fined
Infectious Substance Transport	Explanted Neural Devices	Allegations of transporting unsterilized primate implants (2019).	No Citation in 2024 Report

Operational for Clinical Trials

The confirmation of hazardous material mishandling raises questions regarding the company’s internal safety culture as it from animal research to human clinical trials. The transport of Xylene without proper containment suggests a gap in laboratory safety training for non-clinical staff involved in logistics. also, the reliance on a “fix-it” method, where the fine was reduced because Neuralink agreed to correct the problem retroactively, mirrors the iterative engineering philosophy applied to the device itself. In a regulated medical environment, yet, retroactive compliance is frequently insufficient to mitigate risk. The $2, 480 penalty serves as a formal federal record of non-compliance, establishing a baseline for future regulatory scrutiny as the company expands its physical footprint and logistical complexity in the PRIME study phase.

“We are making a big deal about this because we are concerned for human safety.”
, Internal email from UC Davis employee regarding Neuralink’s transport of explanted devices (2019).

Data Verification Sources

• U. S. Department of Transportation (DOT): Pipeline and Hazardous Materials Safety Administration (PHMSA) investigation records, closed January 2024.
• Reuters: “Musk brain implant company violated US hazardous material transport rules” (Jan 26, 2024).
• Physicians Committee for Responsible Medicine (PCRM): Public records request and complaint filed with DOT (Feb 2023).
• Code of Federal Regulations (CFR): 49 CFR Parts 171-180 (Hazardous Materials Regulations).

SECTION 6 of 22: R1 Surgical Robot Precision: Automated Insertion vs Intraoperative Brain Shift

The “Sewing Machine” method: Micron-Level Insertion

The core of Neuralink’s surgical methodology relies on the R1 Robot, a proprietary automated system designed to insert flexible electrode threads into the cerebral cortex. Unlike traditional deep brain stimulation (DBS) leads, which are rigid and manually advanced, Neuralink’s threads are approximately 5 microns thick, roughly one-tenth the diameter of a human hair, and absence the structural rigidity to penetrate brain tissue independently. The R1 Robot operates on a “sewing machine” principle, using a tungsten needle to grasp a loop at the end of each thread and drive it into the cortical tissue.

The robot use optical coherence tomography (OCT) and computer vision to target specific insertion sites while avoiding surface vasculature. Neuralink claims the system is capable of inserting 64 threads (totaling 1, 024 electrodes) in approximately 20 to 40 minutes. The stated precision of the R1 is in the micron range, theoretically allowing it to place electrodes at specific depths ( 3mm to 5mm) to target 5 of the motor cortex. yet, the static precision of the robot in a controlled environment contrasts sharply with the biological reality of the human brain during and after surgery.

The Brain Shift Phenomenon: Patient 1 Failure Analysis

In January 2024, the implantation of Patient 1 (Noland Arbaugh) exposed a serious disconnect between the robot’s insertion accuracy and post-operative biological mechanics. While the R1 Robot successfully inserted the threads according to the pre-operative plan, the brain is not a static medium. It pulsates with every heartbeat and respiration, and significantly, it shifts when the skull is opened (a phenomenon known as “brain sag” or “brain shift”).

Following the surgery, Arbaugh experienced a retraction of approximately 85% of the implanted threads. Neuralink’s subsequent analysis revealed that the brain had shifted within the skull by up to three times the distance the company’s models had predicted. This excessive movement was exacerbated by pneumocephalus, the trapping of air inside the skull cavity. The “air gap” created space for the brain to move significantly relative to the skull-mounted implant casing. As the brain settled and the air was resorbed, the tethering forces on the threads, which are not anchored to the skull rather float in the brain, caused them to pull out of the cortex, rendering the majority of the electrodes inactive.

Surgical Mitigations: “Skull Sculpting” and CO2 Control

To address the mechanical failures observed in Patient 1, Neuralink implemented specific procedural changes for the second participant, “Alex,” implanted in July 2024. The company introduced a technique referred to as “skull sculpting.” This involves contouring the surface of the skull bone to minimize the gap between the implant housing and the dura mater. By reducing this void, surgeons aimed to eliminate the “air pocket” that allowed for excessive brain motion in the patient.

also, the surgical team implemented strict controls on the patient’s blood carbon dioxide (CO2) levels during the procedure. Elevated CO2 causes cerebral vasodilation, increasing intracranial pressure and brain volume (swelling). By maintaining normocapnia (normal CO2 levels), the team sought to keep the brain volume stable during the insertion process, reducing the likelihood of significant post-operative settling.

Outcome Comparison: Patient 1 vs. Patient 2

The efficacy of these mitigations was confirmed in the months following the second implantation. As of late 2024, Neuralink reported zero thread retraction in Patient 2. The combination of reduced craniectomy gap and tighter physiological control appeared to neutralize the brain shift problem that crippled the device.

Table 6. 1: R1 Robot Insertion & Brain Shift Metrics (2024)

Metric	Patient 1 (Noland Arbaugh)	Patient 2 (Alex)
Implant Date	January 2024	July 2024
Thread Retraction	~85% of threads retracted	0% (No observed retraction)
Primary Cause of Failure	Pneumocephalus (Air Gap) / Brain Shift	N/A
Brain Shift Magnitude	3x expected model	Within expected limits
Key Mitigation	None ( -in-human)	Skull Sculpting, CO2 Control

Adverse Event Reporting: Robot vs. Biological Interface

It is important to distinguish between robot malfunction and biological interface failure. There is no public evidence in FDA reports or Neuralink’s blog updates (up to December 2025) suggesting the R1 Robot itself malfunctioned during the insertion process for either patient. The robot did not “miss” its; rather, the moved post-surgery. The adverse event, loss of data channels, was a result of the device’s design (flexible threads) failing to accommodate the magnitude of post-surgical biological settling.

“In upcoming implants, our plan is to sculpt the surface of the skull very intentionally to minimize the gap under the implant… that put it closer to the brain and eliminate of the tension on the threads.” , Matthew MacDougall, Head of Neurosurgery at Neuralink (July 2024)

The resolution of the retraction problem in Patient 2 suggests that the R1 Robot’s precision is only when paired with rigorous management of the intracranial environment. The “micron-level” accuracy of the robot is rendered irrelevant if the brain moves millimeters or centimeters after the procedure. The 2024 trials demonstrated that the primary challenge for Neuralink is not the robotic insertion itself, the stabilization of the brain-implant interface in the weeks following the surgery.

SECTION 7 of 22: Signal Decoding Algorithms: Software Compensation for Hardware Degradation

Following the mechanical failure of the N1 implant in Patient 1 (Noland Arbaugh), where approximately 85% of the electrode-bearing threads retracted from the motor cortex, Neuralink Corp. was forced to fundamentally restructure its signal processing pipeline. The loss of direct connection to single neurons rendered the initial spike-sorting algorithms, designed to isolate high-fidelity action chance from individual cells, ineffective for maintaining high- cursor control. In response, the company deployed a serious software update in May 2024 that shifted the decoding strategy from single-unit isolation to a broader “neural population signal” analysis.

Algorithm Modification: From Single Spikes to Population

The original decoding architecture of the N1 implant relied heavily on the detection of specific “spikes”, distinct electrical impulses from individual neurons. When the threads retracted, the signal-to-noise ratio degraded, causing these specific spikes to from the recording channels. To compensate, Neuralink engineers modified the recording algorithm to lower the detection threshold, allowing the system to use “neural population signals.”

This shift indicates a transition toward decoding Multi-Unit Activity (MUA) or Local Field chance (LFP), which represent the aggregate electrical activity of a group of neurons rather than a single cell. By aggregating the fainter, distributed signals from the remaining functional electrodes, the updated algorithm could reconstruct the user’s motor intentions even with the sparse input data.

“In response to this change, we modified the recording algorithm to be more sensitive to neural population signals, improved the techniques to translate these signals into cursor movements, and enhanced the user interface.” , Neuralink PRIME Study Progress Update, May 8, 2024

Performance Recovery Metrics: Bits Per Second (BPS)

The efficacy of this software compensation was quantifiable through the standard BCI metric of Bits Per Second (BPS), which measures the speed and accuracy of cursor control. Immediately following the thread retraction in early 2024, Arbaugh’s performance metrics plummeted as the system struggled to decode his intentions with the reduced electrode count.

Post-update data verified by the PRIME study team showed a “rapid and sustained improvement” in performance. While Arbaugh initially set a record of 4. 6 BPS shortly after surgery, the software calibration allowed him to surpass this baseline, eventually achieving peak performances of approximately 8. 0 BPS, and later reported values as high as 9. 51 BPS, even with a significantly reduced channel count. This recovery demonstrated that software adaptability could, to a limited extent, mask catastrophic hardware degradation.

Table 7. 1: Decoding Algorithm Performance Comparison (Patient 1)

Metric	Algorithm V1 (Post-Op)	Hardware Failure State	Algorithm V2 (Post-Mitigation)
Input Source	Single-Unit Spikes (1024 channels)	Degraded/Noise (Retracted threads)	Neural Population Signals (LFP/MUA)
Active Electrodes	~85-90% Functional	~15% Functional	~15% Functional (Optimized)
Cursor Control (BPS)	4. 6 BPS (Initial Record)	Significant Drop (Unpublished low)	8. 0, 9. 5 BPS (Recovered)
Sensitivity Strategy	High-Threshold Spike Sorting	Signal Loss	High-Sensitivity Aggregate Decoding

FDA Compliance and Adverse Event Reporting

The deployment of this software update occurred under the oversight of the FDA’s Investigational Device Exemption (IDE) regulations. While the thread retraction constituted a reportable adverse event due to the mechanical failure of the device interface, the software modification was classified as a mitigation strategy.

Neuralink was not required to seek a new 510(k) or PMA supplement for the algorithm adjustment during the trial phase, as the IDE protocol allows for iterative adjustments to study parameters provided they do not alter the risk profile for the patient. yet, the reliance on software to fix a hardware defect raised scrutiny regarding the long-term viability of the “sewing machine” insertion method. The FDA’s subsequent approval for the second patient (Alex) included a hardware-level change, implanting threads at a depth of 8mm rather than 3-5mm, indicating that while the software patch was for Patient 1, regulators did not accept it as a permanent solution for the mechanical instability of the threads.

Latency and Processing Load

The shift to processing neural population signals inherently changes the computational load on the N1’s onboard ASIC (Application-Specific Integrated Circuit). Decoding aggregate signals frequently requires more complex filtering to distinguish motor intention from background biological noise (such as heartbeat artifacts or vascular pulsation), which the retracted threads were more susceptible to recording. even with these chance latency blocks, the reported increase in BPS suggests that the system’s wireless transmission and external processing units (the “Link” app running on a paired computer) successfully managed the altered data stream without introducing perceptible lag for the user.

SECTION 8 of 22: Thermal Dissipation Safety: Monitoring Cortical Temperature During Inductive Charging

Regulatory Thresholds: The 2°C Limit and ISO 14708 Compliance

The safety of the Neuralink N1 implant during its inductive charging pattern is governed by strict thermal exposure limits set by the FDA and international standards. Specifically, the device must adhere to ISO 14708-1 and ISO 14708-3, which mandate that no outer surface of an active implantable medical device (AIMD) exceeds a temperature of 2°C above the normal surrounding body temperature (37°C) during operation or charging. For the N1, which sits flush with the skull and penetrates the dura mater, maintaining this thermal ceiling is serious to preventing cortical tissue necrosis.

In its 2023 Investigational Device Exemption (IDE) approval process, the FDA raised specific concerns regarding the N1’s lithium battery and the chance for overheating during the wireless power transfer. To secure clearance for the PRIME study, Neuralink was required to demonstrate through rigorous benchtop testing that the device’s inductive coupling efficiency and heat dissipation strategies could maintain the implant-tissue interface 39°C under worst-case charging scenarios.

Inductive Charging Mechanics and Heat Generation

The N1 implant use a resonant inductive coupling system to recharge its custom lithium battery. This process involves an external transmission coil, housed in a wearable “baseball cap” accessory, which generates an alternating magnetic field. This field induces an electric current in the implant’s receiver coil, which is then rectified to charge the battery.

Heat generation during this process from three primary sources:

Heat Source	method	Mitigation Strategy
Joule Heating	Resistive losses in the receiver coil and internal circuitry.	High-efficiency coil design and low-resistance materials.
Eddy Currents	Induced currents in the device’s titanium enclosure.	Ferrite shielding to concentrate magnetic flux away from the case.
Battery Chemistry	Exothermic reactions during the lithium-ion charge pattern.	Strict charge rate limiting (C-rate) and thermal throttling software.

The N1 employs a ferrite shield to direct the magnetic field towards the receiver coil and minimize eddy currents in the hermetic enclosure, which would otherwise act as a heat sink. This design is essential for maintaining charging efficiency while keeping the case temperature within the safe zone.

Real-Time Telemetry and Thermal Throttling

To ensure compliance during the unmonitored home use phase of the PRIME trial, the N1 implant features continuous internal temperature monitoring. The device’s firmware includes a closed-loop thermal management system that actively regulates the charging current based on real-time temperature telemetry.

“Strict regulations govern the charging process to ensure that the temperature doesn’t increase by more than 2 degrees Celsius… a rise in temperature greater than this could damage the brain tissues surrounding the implant.”

If the internal thermistors detect a temperature rise method the safety threshold, the system automatically reduces the power transfer rate or suspends charging entirely. This “thermal throttling” ensures that even if the external charger is misaligned, a common cause of and excess heat in wireless power systems, the implant not overheat the adjacent cortical tissue.

Clinical Observations: Patient 1 and Patient 2

As of early 2026, data from the two human participants, Noland Arbaugh and the second patient (referred to as Alex), indicates that the thermal safety are functioning as designed.

Patient 1 (Noland Arbaugh): Reports indicate that Arbaugh charges his device for approximately 45 minutes to one hour to achieve a full charge, which powers the device for a full day of use. He has described the process as passive and comfortable, wearing the charging cap while relaxing. There have been no public reports of thermal discomfort, headaches, or localized sensations of heat at the implant site, which are early indicators of thermal compliance failure.

Patient 2 (Alex): Following the thread retraction problem seen in Patient 1, Neuralink modified surgical maintained the same core power architecture. Reports from the second implantation confirm that the charging routine remains consistent, with no adverse thermal events recorded in the adverse event logs submitted to the FDA.

Adverse Event Reporting and Long-Term Risks

Under the PRIME study protocol, any instance of the device exceeding the thermal safety limit would be classified as a Serious Adverse Event (SAE) due to the risk of thermal ablation to the brain tissue. As of the most recent verified updates in late 2025, Neuralink has not reported any SAEs related to thermal injury or battery overheating.

yet, the FDA remains vigilant regarding the long-term effects of repeated thermal cycling. While acute overheating is prevented by software controls, the chronic exposure of the meninges and cortex to even mild temperature elevations (e. g., 0. 5°C to 1. 0°C) over years of daily charging is a variable that the 6-year PRIME study is designed to evaluate. The agency requires Neuralink to submit periodic safety reports detailing the thermal history of each implant to ensure that the cumulative thermal dose does not lead to gradual tissue degradation or gliosis around the implant site.

Comparison with Other Active Implants

The N1’s thermal profile is frequently compared to Deep Brain Stimulation (DBS) pacemakers and cochlear implants. yet, unlike DBS pulse generators which are implanted in the chest (where thermal dissipation is more due to blood flow and tissue volume), the N1 is fully integrated into the cranium. This anatomical constraint makes thermal management significantly more challenging, as the skull has lower thermal conductivity than soft tissue, and the brain is highly sensitive to temperature fluctuations.

Neuralink’s successful demonstration of safe inductive charging in this constrained environment represents a significant engineering validation, provided that the long-term data continues to show no histological changes in the tissue surrounding the device.

SECTION 9 of 22: Adverse Event Reporting: Categorizing Thread Retraction as UADE vs Expected SADE

The 85% Signal Loss: Anatomy of a “Known” Failure

In May 2024, Neuralink Corp. publicly acknowledged a significant mechanical failure in its human participant, Noland Arbaugh: approximately 85% of the electrode-bearing threads had retracted from the motor cortex, resulting in a serious loss of bits-per-second (BPS) data transmission. While the company framed this as a solvable engineering challenge, the regulatory classification of this event, specifically whether it constituted an Unanticipated Adverse Device Effect (UADE) or an expected Serious Adverse Device Effect (SADE), carries for the trial’s compliance status.

Under 21 CFR 812. 3(s), a UADE is defined as any serious adverse effect on health or safety that was “not previously identified in nature, severity, or degree of incidence in the investigational plan.” The distinction is serious: a UADE requires immediate reporting (within 10 working days) and an immediate evaluation of whether the trial presents an “unreasonable risk” to subjects, chance triggering a clinical hold. Conversely, a SADE is a known risk that must be monitored does not halt enrollment unless rates exceed projections.

Animal Data vs. Human Reality: The “Unanticipated” Paradox

Investigative reports from May 2024 revealed that Neuralink was aware of the thread retraction risk prior to human implantation. Internal documents and sources by Reuters indicated that similar retraction problem were observed in animal trials involving pigs and monkeys. even with this, the company reportedly deemed the risk sufficiently low to proceed without a redesign of the N1 implant.

This prior knowledge complicates the UADE classification. If Neuralink listed “thread retraction” as a chance risk in its Investigational Device Exemption (IDE) application, the event might technically be classified as a SADE (expected). yet, the severity of the retraction in Arbaugh, where nearly 90% of the electrodes were functionally lost within weeks, likely exceeded the “degree of incidence” projected from animal models. In animal subjects, the brain is smaller and less prone to the magnitude of pulsatile motion (intra-cranial movement due to heartbeat and respiration) seen in humans. Consequently, while the nature of the failure was known, its magnitude in the human cortex was unanticipated.

Regulatory and Mitigation Velocity

The FDA’s decision not to problem a formal clinical hold following the Arbaugh retraction suggests that Neuralink successfully argued the safety profile remained acceptable even with the efficacy loss. The company implemented a dual-track mitigation strategy for the second participant (Alex), implanted in July 2024:

Mitigation Strategy	Technical Implementation	Outcome in Patient 2 (Alex)
Surgical Depth	Threads inserted 8mm deep (vs. 3-5mm in Patient 1) to anchor in cortex.	Zero observed thread retraction as of August 2024.
Skull Contouring	Reduction of the gap between the implant and the brain surface to minimize “tethering” tension.	Stabilized electrode positioning.
Pneumocephalus Control	Strict management of intracranial air (pneumocephalus) which contributed to brain shift in Patient 1.	No significant air trapping reported.

The “Pneumocephalus” Factor

A serious variable in the retraction event was the presence of pneumocephalus, trapped air inside the skull following surgery. In Arbaugh’s case, this air pocket allowed the brain to move more freely within the cranial vault, the tension on the threads. By classifying the retraction as a consequence of surgical technique (pneumocephalus management) rather than a fundamental device flaw, Neuralink likely avoided a Class I recall or a mandated device redesign. This allowed the PRIME study to proceed to the second patient with updated surgical rather than a new device iteration.

Transparency Lag and Public Reporting

While the FDA was likely notified within the mandated 10-day window following the discovery of the retraction in early 2024, public disclosure lagged significantly. Neuralink only confirmed the problem in a May 8, 2024 blog post, months after the surgery and only after inquiries from the Wall Street Journal. This delay in public transparency the between regulatory reporting (which is confidential) and the company’s public narrative of “direct” progress.

For the second participant, Alex, Neuralink reported “no thread retraction” in its August 2024 update, validating the efficacy of the 8mm insertion depth. yet, the reliance on software “over-the-air” updates to boost the sensitivity of the remaining electrodes in Arbaugh’s case highlights a shift in strategy: accepting hardware degradation as a baseline reality and compensating with algorithmic gain.

Investigative Note: The classification of the retraction event relies heavily on the specific wording in Neuralink’s original IDE application. If “migration of threads” was listed as a residual risk, the FDA would view the Arbaugh incident as a validation of that risk rather than a new safety signal, explaining the rapid clearance for subsequent surgeries.

Long-Term Biocompatibility: Gliosis and Encapsulation of Polyimide Threads

Polyimide vs. Silicon: The Biocompatibility Hypothesis

Neuralink’s shift from rigid silicon “Utah Arrays” to flexible polyimide threads was predicated on the hypothesis that matching the mechanical modulus of the implant to brain tissue would minimize the foreign body response (FBR). Standard silicon probes, with a Young’s modulus of approximately 170 GPa, create a significant mechanical mismatch with brain tissue (approx. 3 kPa), leading to chronic micromotion-induced trauma. This trauma triggers the activation of microglia and astrocytes, resulting in a dense sheath of glial scarring (gliosis) that insulates the electrodes and degrades signal quality over time.

The N1 implant use polyimide-insulated gold or platinum-iridium threads, approximately 4 to 6 microns in width, designed to float freely with the brain’s natural movements. Preclinical data from 2015 to 2023 suggested that this flexibility would reduce the “kill zone” of neuronal death around the insertion site. yet, the biological reality observed in the PRIME human trials (2024, 2026) has introduced complex variables not fully accounted for in early animal models. While the flexible design theoretically reduces chronic inflammation, it introduces a new failure mode: the absence of rigid anchoring, which contributed to the thread retraction events observed in early human subjects.

The Retraction Anomaly: Mechanical Failure or Biological Rejection?

In May 2024, Neuralink disclosed that its human participant, Noland Arbaugh, experienced a retraction of 85% of the implanted threads within weeks of surgery. From a biocompatibility standpoint, this event challenged the prevailing assumption that rapid encapsulation would stabilize the threads. In typical intracortical implants, the initial immune response creates a fibrin matrix followed by a glial scar that, while detrimental to electrical conductivity, physically anchors the electrode in place.

The retraction in Patient 1 suggests that the polyimide threads, being ultra-smooth and flexible, did not trigger enough immediate fibrotic response to anchor them against the significant brain shift (up to 5mm) caused by pneumocephalus (trapped air) post-surgery. The threads, tethered to the skull-mounted casing, were pulled out of the cortex as the brain settled. This indicates a paradox in neural engineering: the very property designed to improve long-term biocompatibility (low reactivity/friction) compromised the mechanical stability of the interface in the acute healing phase.

Comparative Analysis: Animal vs. Human Gliosis Rates

Verified data from animal trials (2018, 2023) presents a conflicting picture of long-term biocompatibility. While Neuralink’s marketing emphasized the safety of the threads, FDA inspection records and investigative reports from 2023 revealed significant adverse events in rhesus macaques that were not initially disclosed as biocompatibility failures.

Table 1: Comparative Biocompatibility Findings (Animal vs. Human 2018-2025)

Metric	Rhesus Macaque Trials (2018-2023)	PRIME Human Trial (2024-2025)
Glial Scarring	Moderate; focal reactions at entry points.	Undisclosed/Indirectly inferred via signal stability.
Thread Migration	Not reported as primary failure mode in published data.	85% retraction in Patient 1; mitigated in Patient 2.
Brain Swelling	Severe in specific cases (e. g., “ruptured brain” reports).	Pneumocephalus (air trap) reported; no catastrophic swelling.
Signal Degradation	Variable; implants functional>3 years.	Acute loss due to retraction; remaining channels stable.

Adverse Event Reporting and FDA Compliance

The FDA’s inspection of Neuralink’s California animal facility in June 2023 flagged problem with record-keeping and quality control, specifically regarding the calibration of instruments used to monitor biological responses. This the fidelity of the preclinical data used to justify the safety of the threads in humans. Investigative reports by Wired and Reuters in late 2023 veterinary records showing that monkeys experienced brain swelling and paralysis, which Neuralink attributed to surgical complications rather than the biocompatibility of the threads themselves.

In the human PRIME trial, the “thread retraction” was classified as a device malfunction rather than a biological adverse event (like rejection or infection). yet, the biological method, the brain’s physical movement relative to the implant, is intrinsic to the biocompatibility profile. Neuralink’s subsequent mitigation strategy for Patient 2 (implanted July 2024) involved inserting threads 8mm deep (vs. 3-5mm) to provide more friction and stability. This adjustment implicitly acknowledges that the biological interface requires more mechanical engagement than the original “floating” design provided.

Long-Term Signal Stability and Encapsulation (2025-2026)

As of early 2026, data from the expanded cohort of 21 patients indicates that once the acute stabilization phase is passed, the polyimide threads exhibit stable impedance levels. the chronic gliosis around the threads is indeed less severe than that observed with Utah Arrays. Patient 1, even with the reduced channel count, maintained cursor control accuracy for over 24 months, implying that the remaining 15% of threads were not “walled off” by dense scar tissue, a common fate for silicon probes after one year.

yet, the long-term risk of encapsulation remains. The body’s immune system continuously attacks foreign objects. While polyimide delays this process, it does not stop it. The formation of a high-impedance glial sheath could eventually require higher stimulation voltages or result in the loss of recording resolution (spikes turning into local field chance). Neuralink’s reliance on algorithm updates to decode “neural population signals” rather than single spikes may be a preemptive adaptation to this inevitable biological degradation.

“The challenge is not just getting the threads in, keeping them connected to the same neurons while the brain pulsates, shifts, and tries to scar over the intruder. The retraction event proved that the brain is not a static distinct block of tofu, a organ that fights back mechanically before it fights back chemically.”

Section 11 of 22: Battery pattern Performance: Degradation Rates of the Implanted Power Cell

Inductive Charging Architecture and Thermal Constraints

The N1 implant relies on a custom-designed lithium-ion micro-battery, hermetically sealed within the device’s biocompatible titanium enclosure. Unlike pacemaker batteries designed for low-draw longevity over a decade, the N1 requires daily recharging to support its high- data transmission (1, 024 channels). Power is delivered via a proprietary inductive charging system, integrated into a baseball cap or headpiece worn by the user. This “transcutaneous energy transfer” (TET) system operates by coupling an external transmission coil with the implant’s internal receiving coil, rectifying the induced current to charge the on-board cell.

FDA safety mandates for implantable neuro-stimulators are regarding thermal dissipation. Under the ISO 14708-1 standard, the outer surface of an implant must not raise the temperature of surrounding brain tissue by more than 2°C. For the PRIME trial, Neuralink was required to demonstrate that its fast-charging algorithms could maintain tissue temperature elevations 1°C to prevent thermal necrosis in the cortex. As of January 2026, telemetry data from the 21-patient cohort indicates that the device’s thermal management firmware actively throttles charging currents if internal sensors detect a temperature rise method this safety threshold.

pattern Life Analysis: Patient 1 (Noland Arbaugh)

Noland Arbaugh, implanted in January 2024, represents the longest-running dataset for the N1’s power system. Arbaugh is classified as a “heavy user,” frequently utilizing the BCI for 10 to 12 hours daily for gaming, communication, and productivity tasks. This usage pattern a full charge-discharge pattern nearly every 24 hours. By January 2026, the battery in Patient 1 has undergone approximately 730 complete pattern.

Table 11. 1: Estimated N1 Battery Degradation Metrics (Patient 1 Cohort)

Metric	Jan 2024 (Baseline)	Jan 2025 (12 Months)	Jan 2026 (24 Months)	Status
Runtime (Continuous Use)	11. 5 Hours	10. 2 Hours	9. 1 Hours	Functional
Charge pattern (Cumulative)	0	~365	~730	High Usage
Capacity Retention	100%	~89%	~79%	Degrading
Thermal Delta (Charging)	+0. 4°C	+0. 5°C	+0. 6°C	Compliant

that while the device remains fully operational, the capacity fade is consistent with high-performance lithium-ion chemistries. The reduction in runtime to approximately 9 hours more frequent charging intervals for Patient 1 compared to newer enrollees. Neuralink engineers have mitigated this by optimizing the “sleep mode” power draw, allowing the implant to conserve energy aggressively when no neural activity is being decoded.

Adverse Event Reporting: Power System

A serious focus of the FDA’s pre-market scrutiny involved the risk of battery failure, specifically electrolyte leakage or thermal runaway (explosion) within the skull. Review of the adverse event logs for the PRIME and GB-PRIME (UK) studies reveals:

“No Serious Adverse Events (SAEs) related to battery rupture, electrolyte leakage, or thermal injury have been reported across the 21-subject cohort as of Q1 2026.”

yet, minor anomalies have been noted. Two patients in the expanded 2025 cohort reported “intermittent charging handshakes,” where the external charger failed to couple immediately with the implant, requiring the user to reposition the headpiece. These incidents were attributed to minor swelling at the incision site post-surgery or misalignment of the external coil, rather than internal battery failure. The absence of thermal events validates the efficacy of the N1’s aluminum heat-shielding and the fail-safe circuitry designed to cut power if voltage irregularities are detected.

Long-Term Viability and Replacement

The current degradation trajectory suggests that the N1 battery reach its “End of Service” (EOS) threshold, defined as 60% of original capacity, between years 4 and 5 for heavy users. This lifespan presents a logistical challenge: the N1 is not designed for simple battery swaps. Replacing the power cell requires a full explantation and reimplantation of the device, a procedure that carries surgical risks and the chance for damaging the delicate cortical tissue where the electrode threads are. Neuralink has not yet publicly detailed a “revision protocol” for battery depletion, a regulatory gap that the FDA likely require the company to address before the device receives full commercial Premarket Approval (PMA).

The “Low Risk” Calculation: Pre-Trial Necrosis and Thread Migration

The mechanical failure observed in the PRIME study’s human participant, Noland Arbaugh, specifically the retraction of 85% of the electrode-bearing threads, was publicly characterized by Neuralink as a and unexpected physiological response. Yet, internal veterinary records from the company’s animal testing phase (2017, 2020) and subsequent FDA correspondence indicate that thread migration and cortical damage were known variables long before human enrollment began.

Investigative reports citing former employees and FDA records reveal that the agency’s initial rejection of Neuralink’s human trial application in early 2022 explicitly concerns regarding “micron- threads” migrating to other parts of the brain. even with this regulatory red flag, the company reportedly classified the risk as low, proceeding to human trials without a fundamental redesign of the thread architecture. The correlation between the “unexpected” human retraction and the historical macaque pathology suggests a widespread underestimation of the brain’s viscoelastic response to foreign bodies.

Pathology of “Animal 15”: Cortical Tattering and

Veterinary records released through public records requests by the Physicians Committee for Responsible Medicine (PCRM) provide a direct pathological precedent for the tissue instability seen in human trials. The case of “Animal 15,” a female rhesus macaque implanted in December 2018, documents severe adverse events that mirror the mechanical instability risks flagged by the FDA.

Following the implantation of the N1 prototype, Animal 15 exhibited neurological deficits including “head,” loss of coordination, and shivering. The subject was euthanized in March 2019. The subsequent necropsy report described the brain tissue as “focally tattered” at the implant site, with “remnant electrode threads” found in the cortex. The report further noted “acute cortical necrosis” and evidence of subarachnoid. This “tattering” of the cortex aligns with the shear forces that can occur when a rigid implant (or its tethered threads) does not move in unison with the brain’s natural pulsations, the exact method implicated in the thread retraction observed in the human subject seven years later.

Mechanical Failures: The “Animal 22” Precedent

While the human trial failure was attributed to an “air gap” (pneumocephalus) facilitating thread retraction, similar mechanical decoupling occurred in the primate cohort. “Animal 22” was euthanized in March 2020 after the cranial implant became structurally compromised.

Necropsy data states that the bone screws securing the device to the skull “could easily be lifted out,” rendering the implant “loose.” The report characterized the failure as “purely mechanical.” This loosening of the anchor points creates a where the device floats independently of the skull and brain, the “pistoning” effect that pulls threads out of the neural tissue. In the human trial, the retraction of threads resulted in a loss of; in Animal 22, the mechanical failure necessitated euthanasia due to the impossibility of stabilization.

Comparative Analysis: Macaque Necropsy vs. Human Clinical Events

The following table correlates specific adverse events recorded in the UC Davis macaque cohort (2017, 2020) with the risks and outcomes observed in the PRIME human trials (2024, 2026).

Adverse Event Type	Macaque Subject Data (2017-2020)	Human Clinical Correlation (2024-2026)
Thread Migration	Animal 15: “Remnant electrode threads” found in brain tissue; “focally tattered” cortex.	Patient 1 (Arbaugh): 85% of threads retracted from motor cortex weeks post-op.
Implant Stability	Animal 22: Screws loose; implant “easily pivoted”; mechanical decoupling.	Patient 1: Pneumocephalus (air gap) allowed implant mobility, contributing to retraction.
Infection/Bio-Response	Animal 20: Fungal infection (Candida glabrata); Bio-Glue failure; skin.	General Protocol: Strict sterility required; “Bio-Glue” use modified to prevent cytotoxic compression.
Tissue Necrosis	Animal 15: Acute cortical necrosis; subarachnoid.	Patient 1: No reported necrosis, signal loss indicates electrode displacement from viable neurons.

The Bio-Glue Variable and Granuloma Formation

The use of surgical adhesives, specifically “Bio-Glue,” served as a serious failure point in the animal models that informed the human surgical protocol. In the case of “Animal 20,” the adhesive failed to secure the implant, leading to a fungal infection and subsequent euthanasia in January 2020. More serious, pathology reports from the pig cohort (2021, 2023) revealed the formation of granulomas, inflammatory tissue masses, around the threads.

Reuters reported in 2024 that Neuralink researchers were aware that these granulomas could push the threads out of the tissue or degrade the signal quality. The persistence of this biological rejection response in the animal data contradicts the assertion that the thread retraction in the human patient was an unforeseeable anomaly. The “taming” of the brain, a term used by researchers to describe the subsidence of swelling, was known to alter the relative position of the threads. In the animal models, this frequently resulted in the formation of scar tissue (gliosis) that insulated the electrodes; in the human subject, it manifested as the physical withdrawal of the threads themselves.

FDA 2022 Rejection and the Migration Warning

The FDA’s refusal to approve Neuralink’s initial Investigational Device Exemption (IDE) application in early 2022 was heavily influenced by the animal mortality data. Agency examiners specifically listed “thread migration” as a primary safety concern, questioning whether the device could remain stable in the gelatinous medium of the human brain without causing damage or losing efficacy.

The agency also raised concerns regarding the removability of the device, citing the “tattered” tissue seen in animal explants. While Neuralink eventually secured approval in May 2023 by providing additional safety data, the occurrence of the exact failure mode predicted by the FDA, and evidenced in the necropsy of Animal 15, demonstrates a direct line of causality between the ignored animal pathology and the compromised function in the human trial. The decision to proceed without a hardware redesign placed the load of mitigation on software adjustments rather than solving the underlying mechanical-biological mismatch.

Cranial Integrity: Bone Regrowth Patterns Following Skull Resection

The implantation of the N1 device requires a precise circular craniectomy, removing a coin-sized section of the skull to flush-mount the housing unit. Unlike traditional neurosurgical procedures where bone flaps are replaced, the Neuralink implant acts as a prosthetic skull fragment. As of March 2026, data from the PRIME study indicates that the primary structural challenge is not the failure of the bone-device interface, the management of intracranial volume immediately post-operation.

In the case of the participant, Noland Arbaugh, the surgical site exhibited pneumocephalus, the trapping of air inside the cranial cavity. This atmospheric void prevented the brain tissue from maintaining consistent pressure against the electrode threads. Consequently, the brain moved within the skull during the healing process, contributing to the retraction of nearly 85% of the implanted threads. The “bone regrowth” is functionally replaced by the titanium casing of the N1, which seals the recess. FDA filings confirm that the device relies on this mechanical seal rather than biological osseointegration to maintain cranial integrity.

“The titanium cap seals the recess, restoring skull integrity. The streamlined workflow aims to standardize multicenter deployment.”

For subsequent participants, including the second patient (Alex), Neuralink altered the surgical protocol to mitigate the risk of air entrapment. Surgeons employ techniques to eliminate the air gap between the device and the dura mater before final sealing. This adjustment aims to stabilize the cortex immediately, preventing the “piston-like” motion of the brain that dislodged threads in the initial trial. By March 2026, no serious adverse events (SAEs) related to skull fracture, infection at the resection margin, or device migration have been reported to the FDA, suggesting that the modified cranial seal is even in the absence of natural bone regeneration over the implant site.

Table 1: Cranial Interface Metrics (PRIME Study 2024-2026)

Metric	Patient 1 (Arbaugh)	Patient 2 (Alex)	Status (March 2026)
Resection Type	Circular Craniectomy	Circular Craniectomy	Standardized
Gap Management	Standard Seal	Reduced Air Gap Protocol	Optimized
Complication	Pneumocephalus (Air Trap)	None Reported	Resolved in Protocol
Thread Retraction	High (~85%)	Minimal	Stabilized

SECTION 14 of 22: Data Transmission Latency: Bluetooth Stability in Ambulatory Settings

Wireless Constraints vs. Neural Data Volume

The Neuralink N1 implant operates within a serious engineering bottleneck: transmitting high-fidelity neural data from 1, 024 electrodes through a low-power wireless radio while maintaining a form factor that is hermetically sealed and cosmetically invisible. The device records neural activity at a raw data rate of approximately 200 Megabits per second (Mbps), a volume that vastly exceeds the sustainable throughput of standard Bluetooth Low Energy (BLE) used for ambulatory medical devices. To this gap, the N1 use a custom Application-Specific Integrated Circuit (ASIC) to perform aggressive on-chip data compression and spike detection before transmission.

Unlike wired Brain-Computer Interfaces (BCIs) such as the Utah Array, which stream raw analog signals to external amplifiers, the N1 must process signals internally. The system filters neural “spikes” (action chance) from the background noise and transmits only the digital spike events and cursor commands to the external Neuralink Application. This architecture requires a compression ratio of nearly 200: 1 to fit within the available wireless, which is estimated to be approximately 1 Mbps for stable continuous transmission in a home environment.

Latency Metrics in High-Performance Tasks

The stability and latency of this wireless link were rigorously stress-tested during the 2024 and 2025 PRIME trials, particularly through the subjects’ engagement in high-speed digital tasks.

Table 14. 1: Wireless Latency Performance Indicators (PRIME Study 2024-2025)

Subject	Task	Latency Requirement	Observed Performance	Connection Stability
Patient 1 (Noland Arbaugh)	Mario Kart 8 / Civilization VI	Moderate (approx. 100ms)	Successful navigation; no reported input lag affecting gameplay.	Maintained connection for 8+ hour continuous sessions.
Patient 2 (Alex)	Counter-Strike 2 (FPS)	serious (<50ms)	“Insane” precision; simultaneous movement and aiming without perceptible delay.	Stable during rapid head movements and high data-rate bursts.

Patient 2’s ability to play Counter-Strike 2, a -person shooter requiring millisecond-level reaction times, provides the strongest evidence of the N1’s low-latency performance. While standard Bluetooth mice can introduce latencies of 8-16ms, the N1’s direct neural-to-digital command loop appears to mitigate the mechanical delay of physical muscle activation, chance offsetting the wireless transmission overhead. Alex reported the ability to “look side to side” and aim purely through thought while using a Quadstick for movement, a feat that would be impossible if the wireless link suffered from significant packet loss or jitter.

2. 4 GHz Spectrum Interference and Connection Stability

The N1 radio operates in the 2. 4 GHz Industrial, Scientific, and Medical (ISM) band, a frequency heavily congested by Wi-Fi routers, microwaves, and consumer Bluetooth devices. even with this hostile electromagnetic environment, FDA adverse event reports from the PRIME trial have not radio frequency (RF) interference as a primary cause of failure. The “data loss” incidents reported by Neuralink in May 2024 regarding Patient 1 were explicitly attributed to the mechanical retraction of electrode threads, not a failure of the wireless transmission protocol.

“The N1 wireless interface is capable of transmitting 100, 000 times more data to the computer [than the patient’s BPS output]… The implant is hermetically sealed in a biocompatible enclosure that withstands physiological conditions.”

yet, the reliance on the 2. 4 GHz band presents an ongoing risk for ambulatory patients in uncontrolled environments. To mitigate signal degradation, the N1 employs a custom wideband frequency modulation architecture designed to prioritize signal integrity over range. The range is limited to the immediate vicinity of the user’s computer or mobile device, reducing the likelihood of cross-talk with other devices.

Power Efficiency and Thermal Management

Maintaining a stable high- connection imposes a significant power penalty. The N1’s battery life, demonstrated by Noland Arbaugh’s marathon gaming sessions, indicates a highly radio stack. The implant manages to transmit continuous telemetry for up to 8 hours on a single charge. This efficiency is achieved by offloading the heavy computational load of decoding to the external application, leaving the implant responsible primarily for digitization, spike sorting, and transmission.

The absence of reported thermal events, heating of the implant due to radio transmission, suggests that the specific absorption rate (SAR) remains well within FDA limits for Class III medical devices, even during periods of peak data throughput.

Cybersecurity: Penetration Testing of the N1 Wireless Link

FDA Section 524B and “Cyber Device” Classification

As of March 2026, the regulatory framework governing the Neuralink N1 implant has shifted from general medical device safety to strict “cyber device” compliance under the amended Federal Food, Drug, and Cosmetic Act (FD&C Act). Following the FDA’s issuance of the final guidance “Cybersecurity in Medical Devices: Quality System Considerations and Content of Premarket Submissions” in June 2025, the N1 implant is classified as a Class III cyber device. This designation mandates that Neuralink Corp. maintain a Software Bill of Materials (SBOM) and demonstrate “reasonable assurance” of cybersecurity throughout the device’s lifecycle.

The FDA’s Section 524B requires Neuralink to submit post-market cybersecurity management plans that specifically address vulnerability disclosure and patching timeframes. Unlike earlier Class III devices where software was static, the N1’s reliance on frequent Over-the-Air (OTA) firmware updates to refine neural decoding algorithms places it under continuous regulatory scrutiny. Compliance data from late 2025 indicates that Neuralink has established a dedicated Product Security Incident Response Team (PSIRT) to monitor for vulnerabilities in the N1’s Bluetooth Low Energy (BLE) stack, a serious requirement for maintaining its Investigational Device Exemption (IDE).

Wireless Architecture: BLE Attack Surface and Encryption

The N1 implant transmits neural data wirelessly to an external processing unit (the “Link” app on a smartphone or computer) using a custom profile over Bluetooth Low Energy (BLE). While BLE reduces power consumption, serious for an implant with limited battery capacity, it historically presents a known attack surface, including vulnerabilities such as “Bluesnarfing” or “Knob” attacks that force encryption key renegotiation.

To mitigate these risks, Neuralink employs AES-256 encryption for all data in transit. The pairing process between the N1 implant and the user’s controller use a patented “out-of-band” exchange, requiring physical proximity and a cryptographic handshake that cannot be replicated by standard BLE sniffing tools. This protocol is designed to prevent Man-in-the-Middle (MITM) attacks where an adversary might attempt to intercept motor intention signals or inject malicious commands.

Table 15. 1: N1 Wireless Security Specifications (2025-2026)

Component	Protocol/Standard	Vulnerability Mitigation
Transport	Bluetooth Low Energy (BLE) 5. 3+	Frequency Hopping, Low Energy Secure Connections (LESC)
Encryption	AES-256 (GCM Mode)	Prevents replay attacks and data tampering
Pairing	Proprietary Out-of-Band (OOB)	Physical proximity requirement; rejects standard BLE pairing requests
Firmware Updates	Signed OTA (Ed25519 signatures)	Multi-signature verification prevents malicious code injection

Penetration Testing and Firmware Integrity

Under the 2025 FDA guidance, Neuralink is required to conduct regular penetration testing (“pen testing”) to identify exploit chains that could compromise the device. These tests focus on two catastrophic scenarios: unauthorized stimulation (neuro-modulation attacks) and battery thermal runaway induced by forced continuous transmission.

The N1’s firmware architecture uses a “secure boot” method. Before any code is executed or updated, the on-board processor verifies a digital signature generated by Neuralink’s offline signing keys. If the signature is invalid, indicating a corrupted or malicious update file, the implant rejects the update and reverts to the last known good state. This “anti-bricking” safeguard is essential for a device that requires neurosurgery to replace. Reports from the PRIME trial through December 2025 show no instances of firmware corruption or successful unauthorized access during these update pattern.

Adverse Event Reporting: Cybersecurity Incidents (2024-2025)

A review of the FDA’s Manufacturer and User Facility Device Experience (MAUDE) database and Neuralink’s clinical trial logs reveals zero confirmed cybersecurity breaches or “serious adverse events” (SAEs) related to hacking, unauthorized access, or signal jamming as of January 28, 2026.

“We have observed no evidence of external signal interference or unauthorized device pairing in the 21 subjects currently enrolled across the US, Canada, and UK PRIME study sites.” , Neuralink Clinical Safety Update, January 2026

While mechanical problem such as the thread retraction in Patient 1 (Noland Arbaugh) were widely reported, the digital integrity of the N1 has remained intact. yet, security researchers note that the absence of evidence is not evidence of absence. The expansion of the trial to international sites (GB-PRIME in the UK) introduces new variables, including different cellular interference patterns and chance state-sponsored threat actors interested in BCI data exfiltration.

Data Privacy and the “Telepathy” App

The user interface, branded as “Telepathy,” acts as the gateway for all neural data. To address privacy concerns, Neuralink has implemented a local-processing option. Users can elect to have their neural decoding occur entirely on their local device (smartphone/laptop) without uploading raw neural data to Neuralink’s cloud servers.

For users who opt-in to cloud processing (to improve decoding algorithms), data is anonymized and stripped of personally identifiable information (PII) before transmission. The cloud uplink uses TLS 1. 3 with certificate pinning to prevent interception by compromised network infrastructure. even with these measures, the theoretical risk remains that a compromised smartphone could act as a to attack the implant, a vector that remains a primary focus of ongoing third-party red team operations.

Competitor Benchmarking: Neuralink Invasive Risks vs Synchron Stentrode Safety

The between Neuralink and Synchron in 2026 represents a fundamental split in brain-computer interface (BCI) safety philosophies. While Neuralink pursues a high-, invasive craniotomy method under its PRIME study, Synchron has established a safety baseline with its endovascular Stentrode system. Regulatory filings and trial data from late 2025 and early 2026 indicate a clear contrast in adverse event profiles and procedural risks.

Neuralink PRIME Trial: Invasive Complications and Hardware Instability

Neuralink’s “PRIME” feasibility trial, which expanded to approximately 12 participants by early 2026, continues to grapple with the biological consequences of open-brain surgery. The company’s N1 implant requires a robotic surgeon to remove a portion of the skull and insert 64 flexible threads into the motor cortex. This invasive method has introduced significant mechanical failures. In the case of the human participant, Noland Arbaugh, 85% of the electrode threads retracted from the brain tissue within weeks of surgery. This retraction caused a sharp decline in bits-per-second (BPS) data capture, forcing engineers to rewrite recording algorithms to compensate for the hardware failure.

FDA scrutiny remains high regarding these mechanical vulnerabilities. Reports indicate the agency previously flagged concerns about the implant’s lithium battery safety and the chance for thread migration, risks that materialized in the thread retraction incident. Although Elon Musk announced plans in January 2026 for “high-volume production” and automated surgeries that might bypass dura removal, the current clinical reality involves significant surgical trauma. The risk of pneumocephalus (air trapped in the skull) and the challenge of device removal without damaging brain tissue remain unresolved safety blocks in the 2026 regulatory.

Synchron COMMAND Trial: Zero Serious Adverse Events

In contrast, Synchron’s COMMAND trial data presents a clean safety record. As of the 2026 reporting period, the company documented zero serious adverse events (SAEs) related to the brain or vasculature across its U. S. patient cohort. The Stentrode device avoids the skull entirely, entering the brain’s motor cortex via the jugular vein in a standard endovascular procedure. Clinical data confirms a median deployment time of just 20 minutes, compared to the hours-long complex neurosurgery required for the Neuralink N1.

The COMMAND study met its primary safety endpoint with 100% success. All six U. S. participants retained the device with no permanent increased disability or death. Unlike the thread retraction problem Neuralink, the Stentrode demonstrated long-term signal stability, as the device becomes incorporated into the vessel wall, shielding it from the brain’s rejection response (gliosis) that frequently degrades invasive electrode performance.

Data Visualization: Safety and Procedure Metrics

The following chart compares verified clinical metrics between the Neuralink N1 (PRIME Study) and Synchron Stentrode (COMMAND Study) as of Q1 2026.

Metric	Neuralink N1 (Invasive)	Synchron Stentrode (Endovascular)
Surgical Method	Craniotomy (Skull removal)	Jugular Vein Catheter
Serious Adverse Events (SAEs)	Thread Retraction (85% in Pt 1), Pneumocephalus risk	0% (Brain/Vascular related)
Procedure Time	2-4 Hours (Robotic + Manual)	~20 Minutes (Median)
Signal Stability	Variable (Degraded by retraction)	High (Vessel wall integration)
FDA Status (2026)	Early Feasibility (PRIME)	Pivotal Trial Preparation

Comparative Risk Assessment Chart

Explant Feasibility: Surgical for Device Removal or Upgrade

As of March 2026, the surgical protocol for removing or upgrading the Neuralink N1 implant remains the PRIME study’s most significant unproven variable. While the FDA approved the initial implantation phase, the agency’s original 2022 rejection of Neuralink’s application specifically “unreasonable risk” regarding device removal, noting that the microscopic polymer threads could damage the brain cortex during extraction.

The Stability vs. Removability Trade-Off

Data from the two human participants, Noland Arbaugh (Patient 1) and “Alex” (Patient 2), reveals a serious in surgical strategy that directly impacts explant feasibility. Following the “thread retraction” adverse event in Patient 1, where 85% of the electrode-bearing threads pulled back from the motor cortex, Neuralink altered its surgical protocol for subsequent patients.

Table 1: Surgical Protocol Changes & Explant (2024, 2026)

Parameter	Patient 1 (Arbaugh) Protocol	Patient 2 (Alex) & Subsequent Protocol	Impact on Explant Feasibility
Thread Depth	3mm, 5mm (Cortical Surface)	8mm (Deep Cortex)	High Risk: Deeper insertion increases tissue friction and chance for shearing during removal.
Air Gap Mitigation	Standard Dura Closure	Enhanced CO2 Flush / Skull Sealing	Neutral: Reduces pneumocephalus risk does not alter thread adhesion.
Retraction Rate	High (85% detached)	Reduced (Stabilized via depth)	Inverse Correlation: Greater stability implies stronger tissue integration, complicating clean removal.

FDA Compliance and Adverse Event Reporting

Under 21 CFR 812, Neuralink must report “unanticipated adverse device effects” (UADEs) to the FDA. The retraction problem observed in Patient 1 was classified as a device malfunction that reduced data bitrate, though it did not pose an immediate safety threat requiring emergency explant. Neuralink’s decision to leave the device in situ and modify the recording algorithm established a precedent: software mitigation is preferred over surgical revision.

“The FDA’s primary safety concern regarding explantation involves the chance for the threads to break or scar into the brain tissue, necessitating that they be left behind, a scenario that complicates future upgrades or MRI compatibility.”

The “upgrade” route, replacing an N1 with a future N2 model, requires a clean removal of 64 threads, each thinner than a human hair. Current neurosurgical consensus indicates that after 12 to 24 months of implantation, the brain’s foreign body response (gliosis) encapsulates such electrodes. Removing threads anchored 8mm deep (the Patient 2 protocol) presents a higher risk of cortical tearing than the original 3mm depth. Consequently, the feasibility of a “hardware upgrade” for early PRIME participants remains theoretical, with no successful human explant and replacement procedures reported in the trial data to date.

SECTION 18 of 22: 21 CFR Part 812 Compliance: Investigational Device Exemption Reporting Timelines

Regulatory Mandate: The 10-Day UADE Window

Under the Code of Federal Regulations, specifically 21 CFR 812. 150(b)(1), sponsors of Investigational Device Exemptions (IDE) are legally bound to strict reporting timelines for Unanticipated Adverse Device Effects (UADEs). The statute mandates that a sponsor must submit an evaluation of any UADE to the FDA and all reviewing Institutional Review Boards (IRBs) within 10 working days after receiving notice of the effect. This regulatory clock is serious for patient safety in early feasibility studies like Neuralink’s PRIME trial (NCT06429935), where the device profile is not yet fully characterized in humans.

A UADE is defined as any serious adverse effect on health or safety or any life-threatening problem or death caused by, or associated with, a device, if that effect, problem, or death was not previously identified in nature, severity, or degree of incidence in the investigational plan. For Neuralink, the mechanical failure of the N1 implant in its human subject, Noland Arbaugh, tested the limits of this compliance framework.

The Arbaugh Incident: Discovery vs. Disclosure Gap

On January 28, 2024, Neuralink implanted the N1 device into Arbaugh. According to a company blog post published on May 8, 2024, the retraction of electrode-bearing threads occurred “in the weeks following the surgery.” This vague temporal marker places the adverse event likely in late February or early March 2024. The retraction resulted in a significant loss of data capture, with approximately 85% of the threads detaching from the motor cortex, a failure magnitude that severely compromised the device’s bits-per-second (BPS) performance.

The investigative question centers on the interval between the internal discovery of this retraction and the formal UADE report to the FDA. If the retraction was identified in late February 2024, the 10-day reporting window would have closed in early March. Yet, public acknowledgment and the reported regulatory dialogue regarding a “fix” (implanting threads 8mm deep instead of 3-5mm) did not surface in the public domain until May 2024. While the FDA does not publicly release real-time IDE reports due to commercial confidentiality, the timeline suggests a chance lag between the event’s occurrence and the resolution of the regulatory review that permitted the second patient’s surgery.

Timeline of Events: Patient 1 (Noland Arbaugh)

Date	Event	Regulatory Implication
Jan 28, 2024	N1 Device Implantation	Start of IDE monitoring period.
Feb-Mar 2024	Thread Retraction Discovery	“Weeks following surgery.” Trigger for 10-day UADE report clock.
May 8, 2024	Neuralink Blog Post	Public admission of “thread retraction” and BPS reduction.
May 20, 2024	FDA Clearance for Patient 2	FDA accepts mitigation plan (8mm depth), implying prior review of the adverse event.

The “Anticipated” vs. “Unanticipated” Classification

A serious factor in FDA compliance is whether the thread retraction was truly “unanticipated.” Reuters reported in May 2024 that Neuralink was aware from prior animal testing that the threads could retract. If this risk was documented in the Investigator’s Brochure and risk analysis prior to the human trial, it might technically fall outside the UADE definition, chance exempting it from the expedited 10-day reporting requirement unless the severity (85% loss) exceeded the anticipated risk profile.

yet, if the risk was deemed “low” and not formally characterized as a probable adverse event in the IDE application, its occurrence in the human subject constitutes a UADE. The sheer of the failure, rendering the majority of electrodes useless, for a classification of “unanticipated severity,” which legally the 10-day report. The FDA’s subsequent requirement for a protocol change (increasing insertion depth to 8mm) for the second patient, Alex, confirms that the agency viewed the retraction as a significant safety or effectiveness problem requiring regulatory intervention before the trial could proceed.

Mitigation and Continued Enrollment

Following the review of the adverse event, the FDA permitted Neuralink to proceed with its second participant, Alex, who was implanted in July 2024. The mitigation strategy involved inserting the threads 8mm into the cortex to prevent the “pull-back” effect caused by brain shift and pneumocephalus (air trapped in the skull). As of late 2025, reports indicate that the second patient has not experienced the same degree of retraction, suggesting the mitigation was.

even with this resolution, the opacity of the reporting timeline remains a point of scrutiny. In the broader medical device industry, late reporting is a widespread problem; a 2025 analysis published in the British Medical Journal found that over 1. 2 million adverse event reports were submitted to the FDA after the legal deadline. For a high-profile Class III device like the N1 implant, strict adherence to 21 CFR 812. 150 is not administrative central to validating the safety data that determine future market approval.

Regulatory Note: Under 21 CFR 812. 46(b), if a sponsor determines that a UADE presents an unreasonable risk to subjects, they must terminate the investigation within 5 working days. Neuralink did not terminate the PRIME study rather paused enrollment to implement the surgical modification, a standard pathway when a fix is identified and approved by the FDA.

Internal Safety Culture: Whistleblower Allegations of Rushed Testing Schedules

The “Bomb Strapped to Head” Directive

Internal communications and whistleblower testimony from within Neuralink Corp. reveal a corporate environment where engineering velocity frequently superseded biological safety. Multiple former employees, speaking to investigators in December 2022, described a “pressure cooker” atmosphere driven by Elon Musk’s explicit directive for staff to operate as if they had “a bomb strapped to their heads.” This mandate to accelerate development timelines resulted in what staff characterized as “hack jobs”, hastily executed surgeries that compromised data integrity and necessitated repeat experimentation, so inflating the animal death toll.

The rushed nature of these operations led to preventable surgical errors. In one documented instance from 2021, 25 out of 60 pigs in a single study were implanted with devices of the incorrect size, a logistical failure that rendered the resulting data unusable and required the culling of the subjects. In another case, surgeons implanted the device on the wrong vertebra of two different pigs, leading to severe paralysis and subsequent euthanasia. These errors were not attributed to scientific risks to a absence of preparation time enforced by aggressive internal deadlines.

Animal Welfare Metrics and the 1, 500 Death Toll

Between 2018 and late 2022, Neuralink operations resulted in the deaths of approximately 1, 500 animals, including sheep, pigs, and non-human primates. While animal mortality is a component of pre-clinical medical device testing, whistleblowers alleged that of these deaths were “needless” and driven by the demand to produce headlines rather than scientific validation. The Physicians Committee for Responsible Medicine (PCRM) obtained records indicating that monkeys used in experiments at UC Davis (Neuralink’s partner until 2020) suffered from chronic infections, seizures, and self-mutilation, including animals picking at their implants until they bled.

One particularly contentious incident involved the use of “BioGlue,” a surgical adhesive not approved for the specific procedure in the study protocol. Surgeons applied the adhesive to fill gaps in the skull, which subsequently leaked onto the brain surface of a macaque, causing severe neurological damage and death. Internal documents show that staff had raised concerns about the procedure’s risks were overruled in the push to maintain the testing cadence.

Retaliation and Continued Safety Lapses (2024-2025)

The culture of suppressing dissent extended beyond the animal testing phase into the period of human clinical trials. In June 2024, Lindsay Short, a former animal care lead, filed a lawsuit against Neuralink alleging retaliation after she raised safety concerns. Short claimed she was exposed to the deadly Herpes B virus after being scratched by infected monkeys and was not provided with adequate personal protective equipment (PPE). The lawsuit details that when Short requested better safety and pregnancy accommodations, she was demoted and eventually terminated.

This pattern suggests that the internal safety culture did not materially improve following the 2022 federal investigations. The persistence of basic safety lapses, such as insufficient PPE for handling biohazards, raises serious questions about the rigorousness of the safety applied to the PRIME human trials. The dismissal of FDA reviewers in early 2025, part of broader federal cuts, further reduced the external oversight capacity, chance leaving internal whistleblowers as the primary method for identifying safety deviations.

SEC Investigation into Misleading Safety Claims

In late 2024, the U. S. Securities and Exchange Commission (SEC) reopened an investigation into whether Neuralink and Musk misled investors regarding the cause of animal deaths. Musk had publicly stated on the X platform that “no monkey has died as a result of a Neuralink implant,” claiming the company only used “terminal” monkeys (animals already close to death) for early tests. Veterinary records contradicted this, showing that healthy animals were euthanized specifically due to implant-associated complications, including device failure and infection. This gap between public assertions and internal reality show the widespread opacity that whistleblowers have attempted to expose.

Documented Safety Protocol Deviations (2019-2024)

Date	Incident Type	Outcome	Internal Cause
Aug 2019	Unapproved Adhesive	Macaque death (BioGlue on brain)	Deviation from surgical protocol
Feb 2021	Sizing Error	25 pigs culled	absence of preparation time
Jun 2021	Surgical Error	2 pigs paralyzed (wrong vertebra)	Rushed surgical procedure
Jun 2024	Biohazard Exposure	Staff exposure to Herpes B	insufficient PPE / Retaliation

“In general, we are simply not moving fast enough. It is driving me nuts!” , Internal email from Neuralink executive, in 2022 investigation.

The following is an investigative report section on Neuralink’s PRIME human trial, specifically focusing on patient registry data, long-term follow-up adherence, and FDA compliance regarding adverse event reporting.

Patient Registry Data: Long-Term Follow-Up Adherence in the PRIME Cohort

As of December 31, 2025, Neuralink’s PRIME Study (Precise Robotically IMplanted Brain-Computer InterfacE) has transitioned its initial cohorts from the primary endpoint phase into the serious long-term follow-up (LTFU) period. While the company’s public recruitment registry boasts over 1, 000 registrants with quadriplegia, the active human trial cohort, comprising subjects implanted between January 2024 and late 2025, remains the sole source of verified safety and efficacy data. Adherence to the FDA-mandated 6-year monitoring protocol is the primary metric for assessing the N1 implant’s durability, with early data revealing a between mechanical stability and software-compensated functionality.

PRIME Cohort Retention and Protocol Adherence

The PRIME study protocol (verified under ClinicalTrials. gov ID NCT06429735) mandates a 72-month total duration, split into an 18-month primary study and a 5-year long-term follow-up. Retention data through Q4 2025 indicates 100% subject adherence among the initial implanted cohort, a rarity in early feasibility studies (EFS) for Class III medical devices.

Subject 1 (Noland Arbaugh): Implanted in January 2024, Arbaugh completed his 21-month assessment in October 2025. even with the well-documented retraction of 85% of his electrode threads in mid-2024, Arbaugh has maintained a high adherence rate to the study’s “home use” requirements. The protocol requests a minimum of two 1-hour research sessions per week; yet, telemetry logs indicate Arbaugh frequently exceeds this, utilizing the BCI for 4 to 8 hours daily for tasks ranging from language learning to gaming.

Subject 2 (Alex): Implanted in July 2024, this participant marks a serious control point for surgical mitigation strategies. As of his 15-month check-in (October 2025), Neuralink reported zero thread retraction, validating the efficacy of the revised surgical measures, specifically the reduction of skull sculpting to minimize air gaps and deeper thread insertion (8mm). Alex continues to contribute high-fidelity neural data, using the device for complex 3D modeling (CAD) and competitive gaming, with no reported degradation in signal quality.

Adverse Event Reporting and FDA Compliance

Under 21 CFR 812. 150(b)(1), Neuralink is required to report unanticipated adverse device effects (UADEs) to the FDA and reviewing IRBs within 10 working days. The investigation into the thread retraction problem in Subject 1 revealed a complex compliance. While Neuralink classified the retraction as a “mechanical failure” rather than a “serious adverse event” (SAE), citing no direct harm to the patient, the FDA’s subsequent scrutiny forced a re-evaluation of what constitutes a “reportable malfunction” in BCI threads.

Event Type	Subject	Date Reported	FDA Classification	Mitigation Status
Thread Retraction (85%)	Noland Arbaugh	May 2024	Device Malfunction (UADE)	Software Compensation (Resolved)
Pneumocephalus (Air Gap)	Noland Arbaugh	Jan 2024	Procedural Complication	Surgical Protocol Revision (Resolved)
Signal Degradation	Noland Arbaugh	Feb-Apr 2024	Performance problem	Algorithm Update (Resolved)
Thread Retraction	Alex (Subject 2)	N/A (None)	N/A	Prevented via 8mm Depth

The absence of Serious Adverse Events (SAEs), defined as life-threatening injury, permanent impairment, or death, across the cohort through 2025 is a significant regulatory win. yet, the “thread retraction” incident exposed a gap in the pre-clinical predictive models, which failed to account for the “pumping” motion of the human brain inside the skull, a variable not fully replicable in porcine or non-human primate models.

Long-Term Data Integrity and Registry Expansion

The integrity of the PRIME study’s long-term data relies on the “Neuralink User App,” which streams continuous technical logs to the company’s servers. This telemetry includes electrode impedance, spike detection rates, and Bluetooth connection stability. In late 2025, Neuralink expanded its clinical footprint by initiating the CAN-PRIME study in Canada and the GB-PRIME study in the UK. This international expansion serves a dual purpose: it increases the statistical power of the safety data and tests the device’s compliance with non-US regulatory frameworks (Health Canada and MHRA).

“The transition from acute surgical safety to chronic device durability is the valley of death for BCIs. Neuralink’s ability to keep Subject 1 active and engaged for nearly two years, even with the hardware failure, demonstrates a software resilience that the FDA is weighing heavily against the mechanical risks.”
, Clinical Device Regulatory Analyst, October 2025 Report

By the close of 2025, the patient registry had bifurcated into two streams: the active “implanted” cohort (N=3 confirmed in US, plus international starts) and the “prospective” registry of over 1, 000 screened candidates. The conversion rate from registry to implant remains deliberately low (<1%), driven by inclusion criteria that prioritize stable quadriplegia and a absence of other comorbidities. This selectivity ensures that the long-term follow-up data is not confounded by unrelated health declines, a crucial factor for maintaining the clean safety profile required for future Premarket Approval (PMA) applications.

The following section analyzes the cognitive load and fatigue metrics observed in Neuralink’s PRIME study participants as of March 2026.

Cognitive Load Metrics: User Fatigue During Extended BCI Cursor Control

As of March 2026, the PRIME study has generated significant data regarding the relationship between sustained Brain-Computer Interface (BCI) usage and user fatigue. While traditional BCI research frequently cites “neural fatigue”, the degradation of signal quality or user focus over time, Neuralink’s initial human trials suggest that algorithmic efficiency plays a larger role in endurance than biological exhaustion. The primary metric for assessing this cognitive efficiency is “bits per second” (BPS), a standard unit measuring the speed and accuracy of cursor control.

BPS Performance and Thread Retraction Impact

The experience of the participant, Noland Arbaugh, provided a baseline for understanding how hardware stability dictates cognitive load. Following his January 2024 implantation, Arbaugh initially achieved a control speed of 4. 6 BPS. yet, the retraction of 85% of the electrode-bearing threads in the weeks post-surgery resulted in a severe performance drop to approximately 3. 0 BPS. This mechanical failure forced Arbaugh to exert higher mental effort to achieve the same cursor movements, increasing his cognitive load.

Neuralink engineers responded by modifying the recording algorithm to be more sensitive to neural population signals rather than individual spikes. This software adjustment not only restored function allowed Arbaugh to surpass his initial scores, reaching a peak of 9. 51 BPS by July 2024. This figure method the 10 BPS average of an able-bodied user operating a standard mouse. Crucially, the that once the signal processing was optimized, the “cost” of operation decreased; Arbaugh reported being able to sustain sessions of up to 10 hours on weekends without debilitating fatigue, logging a record 69 hours of usage in a single week.

Transition from Attempted to Imagined Movement

A key factor in mitigating user fatigue has been the transition from “attempted movement” to “imagined movement.” Initially, patients control the cursor by attempting to physically move their paralyzed limbs. Over time, the decoding model adapts to recognize the purely cognitive intent of movement. Arbaugh noted that this shift significantly reduced the mental exertion required for operation. “I no longer have to attempt to move my hand in order to get the cursor to move. I just think that I want the cursor to move and it moves,” he stated in a July 2024 interview. This distinction is important for long-duration viability, as it decouples cursor control from the exhausting neural pathways associated with failed physical motor execution.

Multimodal Control and Efficiency in Patient 2

The second participant, “Alex,” implanted in July 2024, demonstrated how reducing interface friction lowers cognitive overhead. Unlike Arbaugh, Alex experienced no thread retraction due to surgical mitigations that reduced the gap between the implant and the brain surface. Alex utilized the Link to play the -person shooter Counter-Strike 2, combining the BCI for aiming with a mouth-operated Quadstick for movement.

Prior to the implant, Alex relied solely on the Quadstick, which required him to stop moving his character to toggle the joystick into “aiming mode”, a high-friction process that broke his cognitive flow. The addition of the Neuralink allowed for simultaneous movement and aiming. Rather than increasing fatigue by adding a second input stream, the BCI reduced the cognitive load by removing the need for constant mode-switching. Alex described the experience as “,” breaking Arbaugh’s initial records on his day of use. when the BCI functions correctly, it integrates into the user’s motor planning without imposing a “double task” penalty.

Calibration load as a Fatigue Vector

While operative fatigue appears manageable, “calibration load” remains a quantifiable metric of cognitive. Early in the trial, Arbaugh spent up to 45 minutes per session performing recalibration tasks when the decoding model degraded. This maintenance requirement represents a significant non-functional cognitive load. By 2025, Neuralink focused on adaptive algorithms to reduce this time to under five minutes, aiming to eliminate the “start-up cost” that contributes to user frustration and pre-task fatigue.

Table 21. 1: Comparative Cognitive Load Metrics (Arbaugh vs. Alex)

Metric	Noland Arbaugh (Patient 1)	Alex (Patient 2)
Peak Control Speed	9. 51 BPS (July 2024)	>9. 51 BPS (Day 1)
Thread Retraction	85% Loss (High Cognitive )	0% Loss (Nominal )
Primary Fatigue Source	Signal degradation / Recalibration	Game intensity (Normal exertion)
Session Endurance	Up to 10+ hours/day	Variable (Gaming sessions)
Control Method	BCI Only (Cursor)	Hybrid (BCI + Quadstick)

Adverse Event Reporting

As of the March 2026 reporting period, the FDA has not Neuralink for any adverse events related to psychological distress or cognitive burnout in the PRIME study. The absence of “cognitive ” as a reportable serious adverse event (SAE) supports the hypothesis that the device, when properly calibrated, does not impose an unsustainable mental toll. yet, long-term data regarding the cumulative effects of direct cortical stimulation and high- neural data transfer remains a focus of the ongoing feasibility study.

route to Premarket Approval: Statistical Power Analysis for Primary Effectiveness Endpoints

The transition of the PRIME study (NCT06429935) from an Early Feasibility Study (EFS) to a pivotal trial capable of supporting a Premarket Approval (PMA) application hinges on a rigorous statistical demonstration of effectiveness. While the FDA’s 2021 guidance, Implanted Brain-Computer Interface (BCI) Devices for Patients with Paralysis or Amputation, allows for descriptive safety data in early stages, the route to commercialization requires Neuralink to reject the null hypothesis, that the N1 implant offers no statistically significant functional improvement over existing assistive technologies (AT) like eye-trackers or the Quadstick.

Primary Effectiveness Endpoint: The “Bits Per Second” Metric

Neuralink has operationalized “effectiveness” through the quantitative metric of Bits Per Second (BPS), measured via the standardized Webgrid task. This metric combines cursor speed and selection accuracy into a single throughput value. To achieve PMA, Neuralink must demonstrate that the N1 system provides a durable, high- control signal that exceeds the performance of current standard-of-care devices.

Analysis of the initial cohort reveals a serious statistical challenge: variance. In a pivotal trial design, high intra-subject and inter-subject variance the required sample size (N) to achieve a statistical power of 80% or 90% (beta = 0. 1 or 0. 2). Patient 1 (Noland Arbaugh) exhibited extreme variance due to the mechanical failure of thread retraction, while Patient 2 (Alex) demonstrated stability.

Statistical Reality Check: In clinical trials, consistency is as valuable as peak performance. A device that fluctuates between 3 BPS and 9 BPS introduces a standard deviation that complicates the “intent-to-treat” analysis, chance forcing the FDA to require a sample size three to four times larger than originally projected to prove non-inferiority.

Comparative Performance Data: Impact of Thread Retraction on Power

The 85% thread retraction observed in Patient 1 created a “data cliff” that threatened the study’s statistical validity. Prior to the retraction, Arbaugh achieved ~9-10 BPS. Following the mechanical failure, performance plummeted to ~3 BPS before software mitigation strategies restored it to ~8-9 BPS. In contrast, Patient 2, implanted with the mitigated surgical depth of 8mm, showed no retraction and immediate record-breaking performance.

Metric	Patient 1 (Noland)	Patient 2 (Alex)	Standard Assistive Tech (Eye/Mouth)	Able-Bodied Mouse User
Peak Throughput (BPS)	9. 51	>10. 0 (Day 1)	1. 5, 4. 5	~10. 0
Signal Stability (12-Month)	High Variance (Retraction)	Stable	Stable	Stable
Active Electrodes	~15% (Post-Failure)	~100%	N/A	N/A
Statistical Implication	Outlier / High Deviation	Baseline for Pivotal N	Control Arm Benchmark	“Gold Standard” Target

Sample Size Projections for Pivotal Phase

Based on the variance observed in the 21 subjects (as of January 2026), Neuralink’s statistical analysis plan likely a “superiority” claim over existing BCI records (previously ~4. 5 BPS). If the thread retraction problem is definitively resolved, the standard deviation of the treatment effect decreases, allowing for a leaner pivotal trial.

yet, if the “mitigation” strategy (8mm depth) fails to prevent retraction in the broader cohort of 21 patients, the FDA likely mandate a larger sample size to account for the “responder vs. non-responder” dichotomy. The current enrollment acceleration suggests Neuralink is powering the study to detect a mean difference of at least 3. 0 BPS with a p-value <0. 01, a threshold necessary to justify the surgical risk profile of a Class III medical device.

Long-Term Durability and FDA “Leapfrog” Guidance

The FDA’s “Leapfrog Guidance” emphasizes that effectiveness is not a snapshot a longitudinal measure. The agency requires data showing that BPS performance does not degrade significantly over the 12-month primary endpoint window. The “learning effect”, where patients improve BPS through neuroplasticity and software adaptation, serves as a counterweight to signal degradation caused by gliosis (scar tissue formation) or electrode loss.

For the 2026 PMA submission timeline, Neuralink must present a dataset where the slope of performance over time is non-negative. Patient 1’s data, even with the hardware failure, provided a crucial proof-of-concept for software compensation, demonstrating that the ” ” channel count could be lower than 1, 024 if the decoding algorithm is sufficiently strong. This finding may allow Neuralink to for a lower “minimum viable channel count” in their device specifications, reducing the threshold for what constitutes a “device failure” in FDA reporting.