On October 30, 2026, Elon Musk posted on X about Blindsight, the second product from brain-implant company Neuralink. He said it would let "even people who are completely blind, or who have never seen since birth" see. He added that the technology is technically ready and only awaits regulatory approval. But as of November 2, the U.S. clinical trial registry ClinicalTrials.gov lists no Neuralink trial aimed at restoring vision. Neuralink's own website shows only two ongoing clinical trials, one for motor function and one for speech.
There is prior research on implanting electrodes in the human visual cortex. But the results reported so far center on perceiving points of light and simple shapes, and have not reached the level of vision needed for daily life. Earlier visual prostheses have used 60 to 400 electrodes. In the Orion and Utah array studies, which have the most detailed published results, the images people could perceive were limited. Given where this research stands and what Neuralink has made public, Musk's goal of restoring vision is not a demonstrated result today. It should be seen as a future development target.
Musk Says Blindsight Awaits Approval, but No Vision Trial Is Registered
In his October 30 post, Musk claimed Blindsight's technical preparations were complete, but he did not say what approval it is waiting for. Neither Neuralink nor Musk has explained whether this is permission to begin a human clinical trial or some other regulatory procedure. No independent third-party verification of the claim that the technology is ready has been published either.
Checking the public records on ClinicalTrials.gov, seven Neuralink-related trials were registered as of November 2, 2026. All of them target quadriplegia, amyotrophic lateral sclerosis (ALS), or speech impairment, and none concerns vision restoration. Searching for "Blindsight" returns only unrelated studies on hemianopia. The seven registered trials are conducted in the United States, Canada, the United Arab Emirates (UAE), and the United Kingdom, and do not include Japan.
Neuralink's official clinical trial page likewise lists just two ongoing trials: one in which people with quadriplegia use a brain implant to operate devices, and one aimed at restoring speech. There is no listing for a Blindsight trial. The page also states that Neuralink's devices are investigational and have not been approved by regulators such as the U.S. Food and Drug Administration (FDA).
One distinction matters here: the FDA "Breakthrough Device Designation" for Blindsight that Neuralink announced on September 17, 2024. This program supports faster development and review of promising medical devices for serious conditions. It is different from approval, which confirms safety and effectiveness and permits sale. Receiving the designation therefore does not mean Blindsight's clinical effect has been demonstrated.
Public data from animal experiments is also limited. At a conference in June 2025, Neuralink engineer Joseph O'Doherty reported results from monkey experiments. When the visual cortex was electrically stimulated, the monkeys moved their eyes to the location where researchers intended to evoke a visual stimulus in at least two-thirds of trials. This was reported as the first Blindsight experimental result Neuralink had made public, but as of November 2, 2026, no paper or preprint detailing it can be found. Nor can eye movement, a behavioral response, tell us directly what kind of visual experience was occurring.
First Human Implant Slipped from "During 2025" to "Awaiting Approval"
An early specific mention by Musk of Blindsight's development status was a post on X on March 21, 2024. He said Blindsight was "already working in monkeys" and described its initial visual resolution as "like early Nintendo graphics." He has repeatedly commented since then on the potential for vision restoration and the timing of human implantation, but his explanations of the timing have shifted.
The main statements and announcements by Musk and Neuralink, in chronological order, are as follows.
| Date | Speaker | Content |
|---|---|---|
| March 21, 2024 | Musk (X) | Said it was "already working in monkeys." Initial resolution "like early Nintendo graphics" |
| September 17, 2024 | Neuralink (X) | Announced Blindsight received FDA Breakthrough Device Designation |
| September 17, 2024 | Musk (X) | Claimed people blind from birth could see if the visual cortex is undamaged. Initial vision "like Atari graphics" |
| Late March 2025 | Musk (town hall) | Reportedly said he wanted to perform the first human implant during 2025 |
| June 27, 2025 | Neuralink (Summer Update) | Announced a plan to welcome the first Blindsight participant in 2026, aiming to assist mobility |
| January 28, 2026 | Musk (X) | Said preparations for the first implant were ready, subject to regulatory approval. Gave no timing |
| October 30, 2026 | Musk (X) | Said technical preparations were complete and only approval was awaited. Gave no timing |
| November 2, 2026 | ClinicalTrials.gov | No Neuralink vision-restoration trial registered |
Tracing these statements, the outlook for the first human implant changed from "during 2025" in late March 2025 to "2026" in June of that year. From January 2026, no specific timing was given, and the explanation changed to waiting for regulatory approval. As of November 2, 2026, ClinicalTrials.gov still shows no vision-restoration trial corresponding to the plans presented earlier.
In other words, the October 30 statement that technical preparations are complete is not in itself new progress. Musk posted much the same thing on January 28, and even after about nine months, no specific trial start date or details of the approval process have been disclosed. The original explanation, which gave a timeline, has become a conditional one that depends on outside approval.
The ultimate scope of vision restoration Musk envisions also differs greatly from the current state of the technology. In September 2024, he said that even people who had lost both eyes and the optic nerve could see if the visual cortex were intact, and claimed that in the future it might be possible to perceive information at wavelengths such as infrared, ultraviolet, and those used in radar. However, Musk himself acknowledged that initial vision would have very low resolution. In his January 2026 post as well, he projected that it would start with low-resolution vision and improve to high resolution over time.
Earlier Research Achieved Perception of Points of Light and Simple Shapes
In assessing Blindsight's feasibility, key evidence comes from studies that have already implanted electrodes in the human visual cortex. The most detailed published record is the study by Fernández and colleagues, published October 19, 2021, in the Journal of Clinical Investigation. The team implanted a Utah array with 96 electrodes in the visual cortex of a 57-year-old woman who had been totally blind, unable even to sense light, for 16 years. The device stayed in place for six months before being removed, and during that time the researchers examined what perceptions electrical stimulation of the visual cortex produced.
In the experiments, the woman perceived points of light (phosphenes), and with some stimulation patterns she could recognize letter-like shapes. However, she could not identify every letter of the alphabet, and the shapes she could perceive were constrained. The results show that directly stimulating the visual cortex can evoke the perception of light even without visual input from the eyes. But they did not reach a level at which objects or scenery could be recognized as natural images.
Another important earlier case is Orion, a cortical visual prosthesis from the U.S. company Cortigent. The device places 60 electrodes on the surface of the brain and was implanted in six participants between January 2018 and January 2019. According to the company, the small early feasibility study ended in March 2025. When the device was turned on, all participants improved at detecting small squares and motion, and all were rated "positive" or "somewhat positive" in functional assessments.
In Orion, fewer than 4% of electrodes in total lost function, and four of the six participants have had the device removed. These results were presented at the North American Neuromodulation Society meeting in January 2026, but no corresponding peer-reviewed paper can be found at present. Although relatively long-term clinical data have been obtained, the degree of vision restored was limited, and restoration of practical vision has not been demonstrated.
Philip Troyk of the Illinois Institute of Technology, who is involved in developing the 400-electrode cortical visual prosthesis ICVP, also explained as of September 2024 that visual prostheses have limits. He said such devices are currently only an aid to white canes and guide dogs and cannot replace them.
Comparing existing visual prostheses with the technology Neuralink has disclosed gives the following.
| Device / development plan | Electrodes / channels | Electrode placement | Main use | Human track record | Published data |
|---|---|---|---|---|---|
| Orion (Cortigent) | 60 electrodes | Brain surface | Stimulating the visual cortex | Trial with 6 people (2018–2025) | Company conference presentation |
| Utah array (Fernández et al.) | 96 electrodes | Inserted into the brain | Stimulating the visual cortex | 1 person, 6 months | 2021 peer-reviewed paper |
| ICVP | 400 electrodes | Inserted into the brain | Stimulating the visual cortex | Has been implanted in patients | Researcher's explanation |
| N1 (Neuralink) | 1,024 electrodes | 64 threads inserted | Recording neural signals for movement and speech | Used in motor and speech clinical trials | Neuralink official announcements |
| S2 chip (Neuralink) | 1,600+ channels | Not disclosed | Neural stimulation | No human track record disclosed | Neuralink official stream |
| Neuralink's 2026 plan | 3,000 electrodes (target) | Not disclosed | Not disclosed | Planning stage | Neuralink official stream |
Orion, the Utah array, and ICVP, which have reported human visual cortex implants, have 60 to 400 electrodes. The 1,024 electrodes on Neuralink's N1, by contrast, are for reading neural signals related to movement and speech, and are not a track record for Blindsight, which stimulates the visual cortex. As for Neuralink's plan for 3,000 electrodes, no record of use in the human visual cortex can be found as of November 2, 2026. The official specifications, including how many electrodes Blindsight itself will carry, have not been published.
More Electrodes Don't Necessarily Mean Sharper Vision
Electrically stimulating the brain's visual cortex can produce the perception of points of light even when no light is actually entering the eye. A cortical visual prosthesis converts information about the outside world captured by a camera or similar device into electrical stimulation, producing multiple points of light in the visual cortex so that shapes and motion can be recognized. It seems that more electrodes should allow finer images, but the brain's visual processing is not as simple as the pixels of a display.
In July 2024, Ione Fine and Geoffrey Boynton of the University of Washington simulated the perceptions produced by electrical stimulation of the visual cortex and showed that increasing the number of electrodes does not necessarily yield a sharper image. In the visual cortex, thousands of neurons process information in complex ways, and stimulation from an electrode is not perceived as an independent pixel. The research indicated that even with 45,000 electrodes, the perceived image could be quite blurry. Fine pointed out that engineers tend to think of electrodes as pixels, but the brain's visual processing does not work that way.
With actual devices, too, the number of electrodes does not always match the number of usable neural signals. According to a commentary in the Journal of Clinical Investigation on the Fernández study, neural activity could be detected on about one-third of the 96 electrodes in a typical recording. The signals obtained and the effect of stimulation vary with electrode placement and its relationship to surrounding neurons.
Published earlier research alone also cannot establish effectiveness for people blind from birth, as Musk claims. The participant in Fernández's study had visual experience before losing her sight and had been totally blind for 16 years. As for the six Orion participants, the causes of their blindness could not be identified from the materials reviewed this time. It has been shown that if the visual cortex remains, direct stimulation can produce perceived points of light, but whether this applies equally to people blind from birth, and whether they could gain practical vision, is a separate question.
Neuralink's push toward more channels is presumably an effort to send more information by increasing the electrodes that deliver signals to the visual cortex. But more electrodes alone are not what is needed. It also requires technology that understands what perception each stimulus produces and combines them into meaningful images. As earlier research shows, electrode count alone cannot be used to judge visual resolution or usefulness in daily life.
Will a First Demonstration Happen in 2026?
At its Summer Update on June 27, 2025, Neuralink announced a plan to welcome the first Blindsight participant in 2026, aiming to assist mobility. The same stream also presented a concept of increasing the electrode count from 1,000 to 3,000, but it was not made clear whether that number refers to the electrodes Blindsight would carry. With less than two months left in 2026, as of November 2 no Blindsight clinical trial is registered on ClinicalTrials.gov, and no specific schedule for the first human implant has been given.
To judge future progress, the first need is to clarify what the "regulatory approval" Musk refers to actually covers. Beyond that, it will be necessary to check stages such as trial registration, disclosure of trial conditions and target patients, actual human implantation, and verification of results through peer-reviewed papers. Monkeys' eye movements and the perception of light points by earlier devices must be evaluated separately from the restoration of vision usable in daily life.
Even Orion, the leading earlier device, is at the stage of discussing its next large-scale pivotal trial with the FDA, based on clinical data accumulated since 2018. Blindsight may use more electrodes than these earlier efforts, but there is still no public data showing its safety or effectiveness in the human visual cortex. Whether Musk's vision of letting even completely blind people see moves closer to real medical technology depends on whether future human trials can show, with objective data, how much visual function can be restored.