💡 왼쪽 원문을 읽으면서 오른쪽에 따라 써보세요. Tab 키로 힌트를 받을 수 있습니다.

원문 렌더가 준비되기 전까지 텍스트 가이드로 표시합니다.

The BCI industry in 2026 is no longer science fiction. It has been two years and four months since Neuralink implanted the N1 chip in Noland Arbaugh, the first human patient with quadriplegia, in January 2024. In that window the number of invasive-BCI companies has grown from a handful to more than a dozen. At the same time, Apple Vision Pro and the Meta Reality Labs neural wristband have opened a new front of "non-invasive plus consumer" devices. And quietly — Chile and the U.S. state of Colorado have become the first jurisdictions to enshrine "neural rights" in law.

This article maps the 2026 BCI landscape across four axes — invasive, minimally invasive, non-invasive clinical, and consumer / wearable. We cover the technical differences between companies, the data standards (BIDS-iEEG / NWB / HDF5), the analysis software (MNE-Python / FieldTrip / EEGLAB / Brainstorm), and the most recent AI-for-neural-signals research.

1. The 2026 BCI Map — Invasive / Minimally Invasive / Non-Invasive Clinical / Consumer

It is hard to lump BCI together in one sentence because the invasiveness spectrum is enormous. From EEG that just touches the scalp to systems that crack the skull and embed microelectrodes directly into the cortex, the word "BCI" covers a vast range. In 2026, the cleanest split is along four axes.

- Fully invasive (penetrating) — Neuralink N1, Blackrock Utah Array, Paradromics, Precision Neuroscience layer-7

- Minimally invasive (endovascular / subdural) — Synchron Stentrode, Inbrain Neuroelectronics graphene array

- Non-invasive clinical — Onward Medical spinal stimulation, MRI-based brain decoding research

- Consumer / wearable — Emotiv EPOC X, Muse, OpenBCI, Galea, Meta neural wristband, Apple Vision Pro neural tracking

| --- | --- | --- | --- | --- | --- |

Some companies mix axes. Galea integrates OpenBCI (consumer EEG) with a Valve VR HMD, combining EEG + EMG + EDA + EOG + eye tracking in one headset. Cognixion stacks an SSVEP-based BCI on top of AR glasses, letting ALS patients communicate by gazing at flashing characters.

2. Neuralink — First Human Implant (Jan 2024, Noland Arbaugh)

On January 28, 2024, Neuralink implanted its N1 chip into Noland Arbaugh, a quadriplegic patient. The surgery drilled a coin-sized hole in the skull and placed 1024 microelectrodes (64 threads x 16 channels) into the motor cortex. The R1 robot performed the placement. In March of the same year, Arbaugh live-streamed himself playing chess by moving the mouse cursor with his thoughts, and public perception of BCI flipped overnight.

Neuralink's key differentiators are three.

- Wireless charge and wireless telemetry — no external hardware on the head. Bluetooth Low Energy talks straight to a PC.

- Robotic implantation — R1 automatically inserts 0.025 mm thick polyimide threads, far finer than a human neurosurgeon's hand can manage.

- Massive channel count — 1024 channels at launch, an order of magnitude above the Utah Array's 96.

In May 2024, the PRIME study reported that some of Arbaugh's threads had retracted from the brain, losing roughly 85% of the signal. Neuralink compensated in software and recovered performance; the second patient (August 2024) used an improved thread-fixation method. As of late 2025, nine N1 patients have been implanted with more than 4,500 cumulative hours of use.

Conceptual structure of the N1 decoding pipeline as inferred from Neuralink's

PRIME protocol and conference talks. (Actual source is proprietary.)

from scipy.signal import butter, filtfilt

def spike_band_power(raw_voltage, fs=30000, low=300, high=6000):

"""Extract spike-band power from raw 30 kHz samples on the N1."""

b, a = butter(4, [low / (fs / 2), high / (fs / 2)], btype='band')

filtered = filtfilt(b, a, raw_voltage)

return np.mean(filtered ** 2, axis=-1)

def decode_cursor(sbp, weights):

"""2D cursor decoding via a linear Kalman filter."""

return sbp @ weights # shape: (channels,) @ (channels, 2) -> (vx, vy)

3. Synchron — The Stentrode Minimally Invasive Approach

Synchron's Stentrode is the opposite of Neuralink. It never opens the skull. Instead, a catheter is threaded through the jugular vein to park a 16-channel electrode array on top of the motor cortex inside the superior sagittal sinus, balloon-expanded against the wall. The procedure belongs to interventional neuroradiology, not open neurosurgery.

The benefits are twofold.

- Lower risk — equivalent to a stenting procedure. Mortality and infection risk match standard catheterization, not craniotomy.

- Scalability — the existing U.S. interventional-neuroradiology infrastructure (tens of thousands of stroke procedures per year) is immediately usable.

The signal quality lives at the LFP (local field potential) level, not single neuron spikes, and channel count is only 16. Yet intent decoding is fully workable, as 2023 to 2024 trials demonstrated. ALS patients in Australia and the U.S. have been shown texting and emailing, and the FDA granted Breakthrough Device designation in 2024.

Synchron is also the most aggressive early adopter of "intent plus LLM" architectures. The company demonstrated GPT-4 / GPT-5 and Apple Vision Pro integrations long before peers, validating that low channel count plus a large language model can produce surprisingly rich output.

4. Blackrock Neurotech — The Utah Array Research Gold Standard

The Utah Array, originally developed at the University of Utah in the 1990s, is a 96-channel microelectrode array. For more than thirty years it has been the de facto standard for research implants. The 2004 BrainGate trial in which Matt Nagle moved a mouse cursor with his thoughts used the Utah Array, and more than thirty patients since have used the same hardware.

Blackrock's strengths are clinical and academic trust.

- Citations dominate — most invasive-BCI papers use the Utah Array.

- Cereplex / NeuroPort data-acquisition systems are the standard pipeline.

- BrainGate, Pitt, Stanford, and UCSF labs use it as default hardware.

The downsides are visible. The skull must be opened and the chip wired to an external pedestal connector, with infection and cosmetic concerns that prevent daily use outside a hospital. In 2024 Blackrock revealed Neuralace, a wireless fully implantable system, but clinical adoption is set for 2026 to 2027.

5. Paradromics — High Bandwidth

Paradromics, founded in 2015 in Texas, aims at an order-of-magnitude jump in channel count (thousands per device) versus Blackrock. The core idea is a microwire bundle — thousands of thin platinum-iridium microwires tied together and inserted into the cortex.

Paradromics' Connexus Direct Data Interface targets 1600 channels per device, with up to four devices per patient for 6400 total channels. The first FDA Early Feasibility Study approval came in 2025, and the first human implant is scheduled for the second half of 2026. Paradromics is the most direct head-to-head competitor to Neuralink, but its bundle architecture (versus Neuralink's silicon chip) gives a different trade-off in chronic stability versus channel density.

6. Precision Neuroscience — Series C in November 2024

Precision Neuroscience was founded in 2021 by Benjamin Rapoport, a co-founder of Neuralink. Its flagship product, the Layer 7 Cortical Interface, requires only a small skull incision and lays a postage-stamp-sized film electrode on top of the brain. Neither full craniotomy nor penetrating electrodes are involved.

In April 2024, the first temporary human placement happened at Mount Sinai (the array was placed during surgery and then removed). The same November, the company raised a Series C of about 102 million dollars. The plan is to place four 1024-channel films for a total of 4096 channels.

The appeal of Layer 7 is reversibility. Laying it on the brain surface makes removal relatively simple. The downside is that the signal is ECoG (electrocorticography), not single-neuron spikes — one notch below penetrating arrays in spatial resolution. But in clinical-trial velocity and safety, Layer 7 is widely considered the most favorable.

7. Onward Medical — Spinal Stimulation (BCI x SCI)

Onward Medical, based between Switzerland and the Netherlands, makes not pure BCI but a combined BCI plus spinal-stimulation system. The 2023 Nature paper "Walking naturally after spinal cord injury using a brain-spine interface" is the key milestone. Paralyzed patient Gert-Jan Oskam received an ECoG array (brain) and a spinal stimulator (waist) at the same time, and the intent "I want to walk" was transmitted wirelessly from brain to spine, restoring locomotion.

Onward's ARC Therapy received FDA Breakthrough Designation in 2024 and is recruiting patients across five U.S. clinical sites as of May 2026. Among BCI companies, Onward has the clearest "functional recovery (walking)" scenario.

8. Inbrain Neuroelectronics — Graphene Arrays

Inbrain Neuroelectronics is a Barcelona spin-out from ICN2 (Catalan Institute of Nanoscience). Its core technology is graphene-based microelectrodes. Graphene offers two advantages.

- Biocompatibility — immune response is lower than for silicon electrodes, helping chronic stability.

- Optical transparency — compatible with MRI and optogenetics at the same time.

In 2024 the first human surgery took place in Manchester, UK, on a Parkinson's patient. The leading application is the next generation of deep brain stimulation (DBS) — a much thinner graphene lead than Medtronic's existing DBS leads improves stimulation precision.

9. CTRL-Labs (Acquired by Meta 2019) — Reality Labs Neural Wristband

Strictly speaking, CTRL-Labs is not about brain but about wrist EMG (electromyography). Yet it is impossible to leave it out of any "non-invasive neural signal" map, because it is the de facto leader in that lane. Meta (then Facebook) acquired the company for one billion dollars in 2019 and continued R&D quietly inside Reality Labs for six years.

In September 2024 at Meta Connect, the production-ready design was finally unveiled. It is the wrist-worn input for the Orion AR glasses — an EMG band that picks up the micro-movements of fingers to drive virtual clicks, drags, and scrolls. Compared with invasive BCI the resolution is orders of magnitude lower, but the practical edge is decisive — "just put it on, no surgery." It is likely to ship alongside the production version of Orion in 2026.

10. Apple Vision Pro Neural Tracking

The Apple Vision Pro launched in February 2024 using the word "Neural Engine" prominently in its marketing, but technically what it delivers is eye tracking plus hand tracking. You gaze at a target, pinch your fingers, and a click happens. On the surface this is not BCI.

But the "Optic ID" and "Attention" APIs added in visionOS 2 / visionOS 26 are more interesting. They analyze pupil dilation and saccadic patterns to estimate cognitive load and attention. That is essentially an optical BCI that extracts some cognitive signals without EEG. Synchron's direct Vision Pro integration demo sits on top of these APIs.

As of May 2026, rumors place Vision Pro 2 (second half of 2026) with additional sensors. If fNIRS (forehead near-infrared blood-flow measurement) appears in the headband, it would become the first mainstream consumer BCI device.

11. Emotiv + Muse + OpenBCI — The Consumer EEG Trio

Consumer EEG sits at the polar opposite of invasive BCI. Electrodes are placed on the scalp and EEG is recorded. The signal is not medical-grade, but it is good enough for meditation, brainwave visualization, gaming input, and research prototyping. As of 2026 three companies split the market.

- Emotiv EPOC X — 14 channels. After a long history of pseudoscience controversies in the 2010s, the brand is now common in academic research. Its SDK is mature and it is a regular in BCI gaming contests.

- Muse 2 / Muse S — 4-channel meditation headband. Released by InteraXon in 2014, it leads the meditation-app market. Strong in brainwave-driven digital art / music and biofeedback training.

- OpenBCI Cyton + Daisy / Ultracortex — open-source hardware. Extendable to 16 channels. Reasonably priced, with GPL / CERN-OHL firmware that makes it ideal for research and hacking.

Of the three, OpenBCI is the most interesting. Open hardware makes it the easiest path to an EEG prototype at a school or hackathon. The Galea below is the evolution of OpenBCI.

12. Galea (OpenBCI + Valve Cosmos Collaboration) — VR Integration

Galea is a multi-modal biosignal headset built by OpenBCI on top of the Valve Cosmos VR HMD. Announced in 2021, dev kit in 2023, general sale in 2025. A single headset packs the following.

- EEG (10 channels) — scalp brainwave.

- EMG (facial electromyography) — expression detection.

- EOG (electrooculography) — eye motion / blink.

- EDA (electrodermal activity) — arousal / stress.

- PPG (photoplethysmography) — heart rate.

- Eye tracking (eye-facing cameras).

The signal quality is not medical-grade, but the "estimate the user's cognitive and emotional state inside VR / AR" scenario is well within reach. A horror game could adjust difficulty using the user's heart rate, or a meditation app could visualize the user's EEG alpha wave.

Galea also matters as the first serious mass-produced "BCI plus VR" device. Whereas Neuralink targets medical and rehab, Galea targets entertainment and productivity. The price tag is roughly 25,000 to 30,000 dollars, so it is not yet a consumer product.

13. Cognixion — AR Plus BCI

Cognixion is a Santa Barbara company that combines AR glasses with an SSVEP-based BCI (Steady-State Visual Evoked Potential). The core application is communication for ALS patients. ALS robs finger movement quickly, but eye gaze and visual-cortex SSVEP responses remain accessible for longer. The Cognixion ONE projects a character grid on the AR display, and SSVEP detection on the character the patient gazes at translates into typed input.

Compared with invasive BCI from Synchron and others, the signal is noisier, but no surgery is required and a single charge gets you through the day. The device received FDA Breakthrough Device designation in 2024 and insurance reimbursement codes in 2025. Cognixion ONE is widely cited as "one of the first non-invasive BCIs that actually works clinically."

14. Standards — BIDS-iEEG / NWB / HDF5

BCI data is incredibly heterogeneous. EEG, ECoG, spike trains, stimulus metadata, video — all of it can come out of one experiment. Without a shared standard, sharing and reproduction are impossible. As of 2026 the de facto standard is three layers.

- HDF5 — binary file format. Maintained by the HDF Group. Strong on large multidimensional arrays.

- NWB (Neurodata Without Borders) — neuroscience schema on top of HDF5. Driven by the Allen Institute, Kavli, and Janelia.

- BIDS-iEEG / BIDS-EEG / BIDS-MEG — directory-structure standard. Specifies how to lay out files at subject / session / task / run granularity. Driven by the Russ Poldrack lab at Stanford and OHSU.

sub-01/

ses-01/

ieeg/

sub-01_ses-01_task-cursor_run-01_ieeg.edf

sub-01_ses-01_task-cursor_run-01_channels.tsv

sub-01_ses-01_task-cursor_run-01_electrodes.tsv

sub-01_ses-01_task-cursor_run-01_events.tsv

Converting to NWB encapsulates this entire tree inside a single file.

Minimal NWB-file creation

from pynwb import NWBFile, NWBHDF5IO

from datetime import datetime

nwb = NWBFile(

session_description="cursor control task",

identifier="sub-01_ses-01_task-cursor_run-01",

session_start_time=datetime(2026, 5, 16, 10, 0, 0),

)

Fill in ElectricalSeries, Units, and other domain containers, then write.

with NWBHDF5IO("sub-01.nwb", mode="w") as io:

io.write(nwb)

The DANDI Archive (NIH-funded) is the official repository for NWB. As of May 2026 it lists more than 1,500 datasets and over one petabyte of data.

15. Analysis Software — MNE-Python / FieldTrip / EEGLAB / Brainstorm

Roughly 90 percent of the neural-signal analysis market is owned by four open-source tools, each with its own origin and strengths.

- MNE-Python — born at the Martinos Center (Harvard / MGH). Integrated with the Python ecosystem. EEG / MEG / source estimation / connectivity / machine learning in one package. The fastest-growing tool.

- FieldTrip — born at the Donders Institute (Netherlands). MATLAB-based. Richest set of functions for connectivity and time-frequency analysis.

- EEGLAB — born at UCSD SCCN (Scott Makeig). MATLAB-based. Default for ICA decomposition and GUI-friendly workflows. Massive plug-in ecosystem.

- Brainstorm — McGill / USC collaboration. MATLAB-based (free with the MATLAB Compiler runtime). Strongest for MEG / EEG source estimation and visualization.

Extract alpha-band (8-13 Hz) power for one EEG channel using MNE-Python

raw = mne.io.read_raw_edf("subject01_eeg.edf", preload=True)

raw.filter(l_freq=8.0, h_freq=13.0)

psd = raw.compute_psd(method="welch", fmin=1, fmax=40)

alpha_power = psd.get_data(picks="Oz", fmin=8, fmax=13).mean()

print(f"Oz alpha power: {alpha_power:.3e}")

Picking guide — Python ecosystem and ML pipelines, choose MNE-Python; MATLAB lab or ICA-centric, choose EEGLAB; MEG source estimation, choose Brainstorm; connectivity / multi-modal, choose FieldTrip.

16. AI for Neural Signals — Meta Latent Space Brain Decoding / MEG-to-Image (Aug 2023)

The 2023 to 2025 window has produced the fastest answers to the question "what can AI read out of neural signals?" The key papers in short form.

- Meta AI, August 2023 — "Brain decoding: toward real-time reconstruction of visual perception" (Defossez et al.). Reconstructs images seen by a person from MEG signals in near real time (250 ms latency). Training data come from the King et al. 2020 dataset. The trick is mapping brain activity into CLIP embedding space.

- UT Austin, May 2023 — "Semantic reconstruction of continuous language from non-invasive brain recordings" (Tang et al., Nature Neuroscience). Reconstructs the meaning of a story the user listened to from fMRI using GPT-1. fMRI is slow (minute-scale), but it is the first non-invasive semantic reconstruction.

- Princeton / Neuralink, 2024 — Reports on motor-intent decoding accuracy from N1 patients. Kalman filter, then ReFIT (recalibrated feedback intention training), then transformer-based decoder, in sequence.

- 2025 — Meta's follow-up "brain-to-text" research demos MEG-to-text at 1 second latency. Consumer hardware is still far away, but parallel work targets EEG with similar ambitions.

The common pattern is "latent-space brain decoding." Rather than mapping brain signals to characters or pixels directly, align them with the latent space of a pretrained LLM, CLIP, or Diffusion model. Convert what the brain sees into a CLIP embedding and Stable Diffusion can paint the image.

17. The NeuroRights Movement — Chile / Colorado AB1306

As invasive BCI enters clinics and consumer EEG spreads into everyday products, the legal question "who owns your neural data" has surfaced. The NeuroRights Foundation (led by Rafael Yuste of Columbia) is the leading advocacy organization, and two jurisdictions have passed actual law.

- Chile — added a "neural rights" clause to its constitution in 2021, the first country to do so. Mental integrity, cognitive liberty, and mental privacy are codified as fundamental rights.

- Colorado (USA) — 2024 AB1306 (Consumer Protection for Neural Data) classifies individually identifiable neural data as "sensitive personal information" under the Colorado Privacy Act (CPA). Consumer EEG, VR, and fitness-wearable makers feel this directly.

- Minnesota — filed a similar bill in 2025. The EU AI Act is also drafting guidelines to include neural data under "special category."

Five rights are commonly cited — mental privacy, personal identity, free will, equitable access, and protection against algorithmic bias. Full texts are at the NeuroRights Foundation site.

18. Korea — KIST Brain Science Institute / Seoul National University BCI Lab / KAIST plus Samsung

Korean BCI research is stronger in the non-invasive, rehab, and HCI lanes than in invasive systems.

- KIST Brain Science Institute — founded in 1996. The Brain Science Research Division (general BCI) and the Center for Human Enhancement Convergence Research are central. Both invasive animal experiments (monkey BCI) and non-invasive EEG / fNIRS publish heavily.

- Seoul National University BCI Lab — Sung-Jun Kim group (Department of Electrical and Information Engineering). Strong in EEG / ECoG motor-intent decoding and stroke-rehab BCI. Regular top-rankers at the international BCI Competition.

- KAIST Jaeseung Jeong / Department of Brain and Cognitive Sciences — decision-making neuroscience and BCI applications. A "Galaxy plus EEG headband" consumer-BCI collaboration with Samsung Electronics has been reported since 2022.

- Hanyang University Department of Biomedical Engineering / Korea University Department of Brain and Cognitive Engineering — strong in Korean-language SSVEP and P300 spellers.

- Clinical applications — Bundang Seoul National University Hospital and Samsung Medical Center are running BCI rehab trials for stroke patients. A collaboration with Neurable Korea (tentative) was reported in 2025.

On the industry side, LG Electronics unveiled an EEG-based sleep-monitoring prototype out of its healthcare division in 2024, and Kakao Healthcare launched a Muse-integrated meditation service.

19. Japan — Kyoto University / NTT / ATR / RIKEN BSI

Japan is world-class in non-invasive imaging (fMRI / MEG) and cognitive neuroscience.

- Kyoto University — Yukiyasu Kamitani group. The pioneer of fMRI decoding that reconstructs images or even dreams seen by a person. "Neural Decoding of Visual Imagery During Sleep" (Science, 2013) is the landmark.

- ATR Computational Neuroscience Laboratories (Kyoto) — a private research institute founded in the late 1990s. Part of Kamitani's lab infrastructure sits at ATR. Dominant in fMRI brain decoding and neurofeedback.

- NTT Communication Science Laboratories — speech, language, and brain interfaces. A 2024 paper on EEG-based decoding of Korean and Japanese speech intent drew wide attention.

- RIKEN BSI (now CBS, Center for Brain Science) — Japan's national brain-science institute. Invasive monkey BCI and the mapping of single-neuron activity to behavior are the core.

- Keio University — Junichi Ushiba's BCI rehab group. Clinical application of EEG-FES (functional electrical stimulation) systems for hand-function recovery after stroke.

- Industry — Sony unveiled a portable fNIRS-based brain-activity monitor at CES 2024. Panasonic and NEC offer EEG-based fatigue-monitoring solutions for industrial sites.

20. Ethics — Cognitive Liberty, Mental Privacy, Consent

BCI ethics is not just "let's pass a law." The technology creates new gray zones in many places.

- Cognitive liberty — the right of a person to determine their own state of consciousness. What happens when neural stimulation is used by advertisers or political campaigns?

- Mental privacy — who sees your brain data? When the alpha-wave stream sent to the cloud by a consumer EEG ends up sold to insurers, what then?

- Informed consent — did invasive-BCI patients fully understand the "experimental phase" status they signed up for? Who maintains the implant if the company closes five years later (Second Sight's Argus II retinal implant in the early 2020s already lived this).

- Algorithmic bias — BCI decoders inherit biases of their training data. Does a decoder trained on white males maintain accuracy for women, Asian patients, or the elderly?

- Free will — if neural stimulation directly drives behavior, who is responsible for that behavior?

These five questions are the spine of the late-2020s NeuroRights legislative debate and the regions partially answered by Chile and Colorado. The full answers, however, are nowhere near complete.

21. Who Should Learn BCI — Medicine / CV / VR / Neuroengineering Scenarios

Closing with a scenario-by-scenario guide to "should I study BCI?"

- Medicine and rehab — follow the Onward Medical / Synchron / Blackrock track. Learn MNE-Python plus BIDS plus NWB, and study clinical IRB and medical-device approval (FDA / MFDS).

- Computer vision and deep learning — reproduce Meta Latent Space Brain Decoding or MEG-to-image. fMRI has rich public datasets (Human Connectome Project, NSD); MEG has the King et al. 2020 dataset as a good starting point.

- VR / AR / games — learn Galea, the Apple Vision Pro neural-tracking API, and the Meta neural wristband. Consumer BCI is low accuracy but it produces new UX paradigms in games, productivity, and accessibility.

- Neuroengineering and hardware — build your own EEG headset with OpenBCI Cyton plus Ultracortex, then look at the hiring pages of Paradromics and Precision Neuroscience. Electrode materials (Pt-Ir, graphene, polyimide) and ASIC design are the core skills.

- Law, ethics, and policy — read the NeuroRights Foundation, the IEEE Brain Initiative, and the OECD Recommendation on Neurotechnology. Colorado AB1306 is the model text.

- General developer — pick up a Muse or an OpenBCI kit and build a weekend project. EEG-driven music, meditation visualizations, mind-controlled mini games — these prototypes have the lowest barrier to entry.

Two big principles. First, the invasiveness spectrum is too wide to lump under the single word "BCI." Second, the standards (BIDS / NWB) and the analysis tools (MNE-Python) are the common substrate in every scenario, so learn those first.

22. References

- Neuralink PRIME Study — https://neuralink.com/blog/prime-study-progress-update/

- Noland Arbaugh first livestream — https://x.com/neuralink/status/1770563939413496146

- Synchron — https://synchron.com/

- Stentrode SWITCH Trial — https://clinicaltrials.gov/study/NCT03834857

- Blackrock Neurotech — https://blackrockneurotech.com/

- Utah Array (BrainGate) — https://www.braingate.org/

- Paradromics — https://paradromics.com/

- Precision Neuroscience — https://precisionneuro.io/

- Layer 7 first human use — https://www.mountsinai.org/about/newsroom/2024/precision-neuroscience-completes-historic-mount-sinai-trial

- Onward Medical — https://onwd.com/

- Brain-spine interface (Nature 2023) — https://www.nature.com/articles/s41586-023-06094-5

- Inbrain Neuroelectronics — https://www.inbrain-neuroelectronics.com/

- CTRL-Labs / Meta Reality Labs neural wristband — https://about.fb.com/news/2024/09/meta-orion-true-augmented-reality-glasses/

- Apple Vision Pro — https://www.apple.com/vision-pro/

- visionOS Attention API — https://developer.apple.com/documentation/visionos/

- Emotiv — https://www.emotiv.com/

- Muse — https://choosemuse.com/

- OpenBCI — https://openbci.com/

- Galea — https://galea.co/

- Cognixion — https://www.cognixion.com/

- BIDS-iEEG — https://bids-specification.readthedocs.io/en/stable/modality-specific-files/intracranial-electroencephalography.html

- BIDS-EEG — https://bids-specification.readthedocs.io/en/stable/modality-specific-files/electroencephalography.html

- NWB (Neurodata Without Borders) — https://www.nwb.org/

- DANDI Archive — https://dandiarchive.org/

- MNE-Python — https://mne.tools/stable/index.html

- FieldTrip — https://www.fieldtriptoolbox.org/

- EEGLAB — https://sccn.ucsd.edu/eeglab/index.php

- Brainstorm — https://neuroimage.usc.edu/brainstorm/

- Meta Brain Decoding (2023) — https://ai.meta.com/blog/brain-ai-image-decoding-meg-magnetoencephalography/

- MEG-to-image (Defossez et al., arXiv 2023) — https://arxiv.org/abs/2310.19812

- Semantic reconstruction (Tang et al., Nature Neuroscience 2023) — https://www.nature.com/articles/s41593-023-01304-9

- NeuroRights Foundation — https://neurorightsfoundation.org/

- Chile constitutional neural rights — https://www.bcn.cl/leychile/navegar?idNorma=1166983

- Colorado AB1306 — https://leg.colorado.gov/bills/hb24-1058

- IEEE Brain — https://brain.ieee.org/

- KIST Brain Science Institute — https://www.kist.re.kr/

- SNU BCI Lab — https://bcilab.snu.ac.kr/

- Kyoto University Kamitani Lab — https://kamitani-lab.ist.i.kyoto-u.ac.jp/

- ATR Computational Neuroscience Labs — https://www.cns.atr.jp/

- RIKEN CBS — https://cbs.riken.jp/en/

- BCI Competition — https://www.bbci.de/competition/