Originally published on The Daily Flare.
Brain computer interfaces are moving from lab experiments to real medical trials, restoring communication, mobility, and independence to people with severe paralysis.
Brain computer interfaces — devices that translate neural signals directly into digital commands — have crossed a line in recent years. What began as a research curiosity is now a clinical reality: participants in trials are controlling cursors, typing sentences, operating robotic arms, and even speaking through synthesized voice, all by thought. As of October 2026, no implantable BCI has yet received full commercial approval in the US or EU, but the field is closer than it has ever been. This tracker brings together the verified milestones, the key companies, what the technology can actually do today, and the honest challenges that remain.
What Brain Computer Interfaces Are
A brain computer interface creates a direct communication path between the brain and an external device. Electrodes record the electrical activity of neurons; software decoders translate those patterns into actions — moving a cursor, selecting a letter, or driving a robotic arm.
Approaches differ sharply in invasiveness. Intracortical systems (Neuralink's threads, Blackrock's arrays) penetrate the cortex for high-bandwidth signals but require open-brain surgery. Minimally invasive options — Synchron's Stentrode, threaded through a blood vessel with no skull opening, or Precision Neuroscience's thin-film array laid on the brain surface — trade some signal fidelity for lower surgical risk. Non-invasive EEG headsets record through the skull at zero medical risk, but the skull smears the signal, so they are limited to simpler tasks like focus and wellness tracking.
Brain Computer Interfaces Advancement Timeline: 2023 to 2026
The pace of the last three years has been extraordinary. Here is what actually happened, in order:
- May 2023: The US Food and Drug Administration grants Neuralink approval to begin first-in-human clinical studies of its N1 implant, opening the door to human trials of the high-bandwidth intracortical approach.
- September 2024: Synchron reports 12-month COMMAND study results: across six implanted patients, zero brain- or vasculature-related serious adverse events and 100% device deployment success, with a median deployment time of just 20 minutes.
- September 2024: The FDA grants Neuralink's Blindsight vision-restoration device Breakthrough Device Designation — though as of September 2026 no human implant has occurred and no trial is registered.
- November 2024: Neuralink launches CONVOY, a trial of thought-controlled robotic arm use, expanding beyond cursor control into physical assistance.
- April 2025: Precision Neuroscience receives FDA 510(k) clearance for its wireless Layer 7 system — the first full clearance for a wireless BCI — for implantation of up to 30 days in temporary clinical use.
- October 2025: Neuralink launches VOICE, a speech-restoration trial, and demonstrates a participant feeding himself a pretzel with a thought-controlled wheelchair-mounted robotic arm.
- November 2025: Paradromics is granted an FDA Investigational Device Exemption for speech restoration with a fully implantable BCI — a regulatory first — and Synchron raises $200 million in Series D funding for its pivotal trial and first-generation commercial launch.
- January 2026: Neuralink's "Two Years of Telepathy" update confirms 21 participants worldwide, zero serious device-related adverse events, and a roadmap toward next-generation hardware with roughly 3,000 electrodes. Precision partners with Medtronic.
- March 2026: BrainGate's bimanual QWERTY decoding study, published in Nature Neuroscience, sets a typing record of 110 characters per minute at a 1.6% word error rate.
- May 2026: Neuralink completes its first transdural implant in Canada, a less invasive surgical approach.
- June 2026: Paradromics completes its first permanent human implant at University of Michigan Health in a woman with motor neuron disease, training the system on thousands of words and sentences. Separately, a UC Davis study published in Nature Medicine documents ALS patient Casey Harrell using an intracortical speech BCI independently at home for nearly two years.
- October 2026: Neuralink reports an engineering update: 50,000+ hours of unlabeled neural data used to pretrain decoders, cutting calibration from roughly 10 minutes per day to roughly 10 minutes per week for some users, plus a new 11.32 bits-per-second cursor record. (Both are company-blog claims, pending peer review.)
The Key Players: A Comparison
| Company | Approach | Electrodes | Humans implanted | Regulatory status |
|---|---|---|---|---|
| Neuralink | Intracortical penetrating threads (N1) | 1,024 (~3,000 planned) | 21 confirmed (Jan 2026) | IDE trials (PRIME, CONVOY, VOICE); Blindsight has Breakthrough Designation but no human implant |
| Synchron | Endovascular Stentrode via blood vessel (no open-brain surgery) | 16 | 10 | Breakthrough Device Designation + IDE; preparing 2026 pivotal trial toward first BCI premarket approval |
| Paradromics | Fully implanted Connexus (speech-focused) | 400+ | 1 (first permanent implant, Jun 2026) | FDA IDE (Nov 2025) — first for fully implantable speech BCI |
| Precision Neuroscience | Layer 7 thin-film array on brain surface | 4,096 (across four arrays, May 2024) | 37+ tested | FDA 510(k) clearance (Apr 2025) for up to 30-day temporary use |
| Blackrock Neurotech | Utah-array-style intracortical implants | Varies | ~36 (research-grade) | Research use |
| Neurable | Non-invasive EEG headphones | N/A (EEG) | N/A | Consumer wellness device |
| Merge Labs | Not yet disclosed | — | 0 | New entrant (raised $252M in 2026) |
What this table shows
No single design is winning outright — the field is splitting into lanes. Neuralink and Paradromics chase the highest bandwidth with surgically implanted electrodes. Synchron trades bandwidth for safety, threading its device through a blood vessel in about 20 minutes with no skull opening at all. Precision sits in between with a high-electrode-count film on the brain surface and the only FDA clearance of any kind — though strictly for temporary, 30-day clinical use. Notably, no implantable BCI is commercially approved in the US or EU as of October 2026; every device listed is investigational.
What BCIs Can Actually Do in 2026
Cut through the headlines and these are the verified, documented capabilities:
- Cursor control: Neuralink participants routinely control computer cursors by thought — browsing, emailing, gaming. One participant exceeded 10 bits per second, matching or exceeding able-bodied mouse control (8–10 BPS). A new 11.32 BPS record was claimed on the company blog in October 2026 but is still pending peer review.
- Typing: BrainGate's 2026 study demonstrated 110 characters per minute — 22 words per minute — with a 1.6% word error rate, calibrated from as few as 30 sentences and used at home.
- Speech: The most moving milestone of 2026. ALS patient Casey Harrell used an intracortical speech BCI independently at home for nearly two years — more than 3,800 hours and 1.96 million words — at an average of 56 words per minute, with over 99% word accuracy on large-vocabulary tasks. Peer-reviewed in Nature Medicine on June 15, 2026.
- Robotic arm and mobility: Neuralink's CONVOY program demonstrated a participant feeding himself a pretzel with a thought-controlled wheelchair-mounted arm; thought-controlled powered wheelchairs followed in July 2026.
- Creative expression: A 69-year-old quadriplegic participant in Blackrock's research program composed music from neural signals in 2026.
What they cannot do: there is no true "consumer thought control." Non-invasive headsets can track focus and wellness, but consumer marketing that implies mind-reading or telepathy-style control is overpromising. Every serious capability listed above requires implanted electrodes — and, for now, a clinical trial.
Invasive vs Non-Invasive: The Core Tradeoff
The entire field is organized around one tradeoff: signal quality versus surgical risk.
Invasive BCIs — Neuralink's threads, Paradromics' Connexus, Blackrock's arrays, Precision's surface film — record from dozens to thousands of electrodes placed in or on the brain. That density is what makes 11+ bits-per-second cursor control and 22–56 words-per-minute decoding possible. The cost is neurosurgery, with all its attendant risks, and the price of hardware built largely by hand.
Minimally invasive BCIs — Synchron's Stentrode through a blood vessel, Precision's thin film through a sub-millimeter incision — aim for the middle ground: real neural recordings with dramatically lower surgical burden. Synchron's median deployment time of 20 minutes and clean 12-month safety data make this lane the closest to first commercial approval.
Non-invasive BCIs — EEG headsets like Neurable's — carry zero medical risk. But the skull smears electrical signals, limiting resolution. They are legitimate wellness and focus tools, not medical-grade communication devices. Anyone selling EEG headbands as "thought control" is selling science fiction.
What's Next
- Synchron's pivotal trial: Expected in 2026, required before the company can seek the first premarket approval of a permanently implanted BCI. If it succeeds, history gets made: the first FDA-approved implantable brain computer interface.
- Neuralink's scale-up: High-volume N1 production at an Austin factory, automated R1 surgical robot, and next-generation hardware with roughly 3,000 electrodes. The Blindsight vision program targets a first human implant in the second half of 2026 — a stated intention, awaiting final regulatory approval, with no human implant yet as of September 2026.
- Paradromics' speech push: Enrolling more participants in the Connect-One study, with a six-year follow-up plan aimed at closed-loop, real-time text and synthesized voice.
- Precision's path to permanence: Its $250 million Series D and Medtronic partnership are funding a push beyond the current 30-day clearance toward permanent implants.
Honest Challenges: What Still Has to Be Solved
- Signal longevity. Nobody knows how well these devices perform beyond two to five years. Electrodes scar, signals drift, and lifelong implants need decades of data that don't exist yet.
- Calibration burden. Decoders must learn each user's neural patterns. Neuralink's claim that pretraining on 50,000+ hours of neural data cuts calibration from 10 minutes daily to 10 minutes weekly is promising — but it is company-reported, not peer-reviewed.
- Bandwidth versus safety. Higher electrode counts decode more, but penetrating the brain carries more risk. There is no free lunch, and every design is a bet on where the balance settles.
- Cost and manufacturing. Today's devices are largely hand-built. Affordable, mass-manufactured implants are a prerequisite for reaching the millions of people with paralysis — not just dozens of trial participants.
- Ethics and privacy. Neural data is the most intimate data there is. Ownership, lifelong consent, data security, and equitable access are unresolved — and unlike a phone, you can't just stop using your brain.
- Evidence quality. No peer-reviewed human data from any Neuralink device exists as of October 2026. The field's headline records — 11.32 bits per second, calibration improvements — come from company blogs. The strongest published evidence, like Casey Harrell's 3,800 hours of independent at-home speech decoding, comes from academic research teams. That gap between company claims and peer-reviewed results is worth remembering every time a new record is announced.
The Bottom Line
Brain computer interfaces have left the laboratory. People with paralysis are already communicating, computing, and reaching into the physical world by thought — the milestones of 2024 through 2026 are real and documented. But the technology remains investigational: no device is commercially approved, the longest-term safety data spans only a few years, and the most dramatic claims still await independent verification. The coming years — Synchron's pivotal trial, Neuralink's scale-up, Paradromics' speech work — will determine whether BCIs become a medical standard or remain a spectacular promise.
Top comments (1)
Việc chuyển dịch từ môi trường phòng thí nghiệm sang ứng dụng thực tế như bạn đề cập thực sự đặt ra những thách thức cực lớn về độ trễ tín hiệu. Khi chúng ta nói về điều khiển thiết bị qua ý nghĩ, chỉ cần một vài mil giây chênh lệch giữa tín hiệu thần kinh và phản hồi của phần mềm cũng đủ để gây ra cảm giác mất kết nối hoặc thậm chí là chóng mặt cho người dùng. Tôi thấy giới nghiên cứu đang tập trung rất nhiều vào việc xử lý nhiễu tín hiệu bằng các mô hình học máy ngay tại edge để giảm tải cho việc truyền dữ liệu về trung tâm. Nếu giải quyết được bài toán xử lý tại chỗ này, tốc độ phản hồi sẽ đạt đến mức gần như tức thời.