Elon Musk’s brain-to-computer venture Neuralink has its second product on the launchpad, and it’s a doozy.
In a post on X, Musk said Blindsight would enable people who are completely blind or have never been able to see to gain sight.
“It is ready from an engineering standpoint,” he wrote. “Just waiting on regulatory approval.”
Our second product is called Blindsight and it will, as the name suggests, enable even those who are completely blind or have never seen to see. It is ready from an engineering standpoint. Just waiting on regulatory approval.
— Elon Musk (@elonmusk) October 30, 2026
Neuralink was founded in 2016. Its first product, Telepathy, enables high-speed communication with computers or humans simply by thinking, according to Musk.
While much brain-computer interface (BCI) research has focused on people controlling objects with their brain, Blindsight uses a camera to produce visual information.
That information is converted into stimulation instructions and sent to a device implanted in the brain, which interprets the stimulation as visual sensations.
The sensations are perceived as spots or flashes of light called phosphenes. By driving thousands of these points simultaneously, the brain constructs a real-time, low-resolution visual map of a person’s surroundings.
The result is synthetic vision, which initially will be coarse and low-resolution, rather than normal eyesight.
Expect Regulatory Delays
Although Blindsight may be ready from an engineering standpoint, it will likely be waiting around for regulatory approval for some time.
“There’s a big gap between ‘the device works on the bench’ and ‘the FDA will let you put it in people’s brains,'” declared Fylaktis Fylaktou, a postdoctoral researcher at UCLA’s Semel Institute for Neuroscience and Human Behavior.
“A cortical visual prosthesis is a Class III device, which is the most heavily regulated category we have,” he told TechNewsWorld. “Approval requires staged safety trials over several years.”
“Musk’s announcement tells you Neuralink has manufacturing confidence in their hardware, but the regulatory timeline is another problem,” he said.
“The basic method isn’t new,” he continued. He pointed out that William H. Dobelle, a biomedical researcher at the University of Utah and pioneer in artificial vision, demonstrated phosphene generation from cortical stimulation back in the 1970s.
“What Neuralink brings is higher electrode density and a surgical robot that makes implantation scalable,” he explained. “That’s engineering progress, not a scientific breakthrough.”
“What patients will actually perceive is ultra-low-resolution patterned light, closer to a dot-matrix display than vision,” he said.
“In people who have been blind since birth, the visual cortex has typically been recruited for other functions like touch, hearing, and language,” he added. “Stimulating it may not produce visual experience at all.”
Flashes of Light Not Enough
Daniel Palanker, a professor of ophthalmology at Stanford University’s School of Medicine and a researcher of optical and electronic technologies for diagnostic, therapeutic, surgical and prosthetic applications, noted that two studies with cortical stimulation for restoration of sight in human patients demonstrated a major problem with the approach.
“Pattern stimulation does not result in a corresponding patterned perception,” he told TechNewsWorld. “Electrodes can elicit perception of flashing light — phosphene — but several electrodes activated in a pattern do not produce a perception of a superposition of these phosphenes from its component electrodes.”
“Rather, it produces yet another phosphene of a different shape, size, and location,” he continued. “Therefore, so far, cortical stimulation did not yield any useful vision beyond just the sensation of a flickering light.”
“One workaround for this problem was to slowly move stimulation from one electrode to another — about one-half second per electrode, creating a sensation of a slowly moving single spot of light, but this approach is not scalable to a large number of electrodes and to a video frame rate,” he said.
“It will be interesting to see whether Blindsight will yield different results in human patients,” he added.
Implant Durability and Patient Risks
Deploying Blindsight will have its challenges. “Chronic electrode viability is the hard one,” Fylaktou noted. “Glial scarring and signal drift degrade implants over months to years, and nobody has fully solved it.”
“Resolution is another,” he continued. “Musk himself has mentioned ‘Atari graphics’ as the starting point, which is honest, but whether it will be functional remains to be seen.”
“Cortical plasticity in long-blind patients compounds both,” he added.
While intended for good, Blindsight can pose harmful risks. “Cortical implantation carries surgical risk, including infection, hemorrhage, and seizure,” Fylaktou explained.
“Chronic inflammation and device failure can require explant surgery, which is its own trauma,” he continued.
“There’s also a psychological harm pattern that showed up in earlier BCI trials when patients recruited with ‘restore your sight’ framing experienced real distress when the perceptual experience turned out to be nothing like vision,” he added.
“Informed consent for first-in-human cortical implants has to be unusually careful about managing expectations,” he warned.
Sense and Stimulation
Ranya Belmaachi, product lead at Subsense, in Palo Alto, Calif., a developer of a non-surgical brain-computer interface, noted that restoring lost senses is a natural first target for BCIs, because decades of research have created a deep understanding of how our senses work. “In fact, you could argue the cochlear implant was the first successful BCI,” she told TechNewsWorld.
“Beyond the senses,” she continued, “today’s BCIs are focused on reading motor and speech intent to help people with paralysis or locked-in syndrome regain their independence.”
“But BCIs don’t just read from the brain,” she added. “They can also stimulate it, which is called neuromodulation.”
“That opens the door to treating neurological conditions like Parkinson’s disease, supporting stroke rehabilitation, preventing seizures, and even addressing psychiatric conditions like treatment-resistant depression,” she said.
“The brain is at the center of our health and being able to both read from it and stimulate it will help us understand and treat it in novel ways,” she observed.
“The next challenge is access,” she maintained. “At Subsense, we’re working on a non-surgical approach so these benefits can reach far more patients, much earlier in their disease progression.”
Musk’s Influence Cuts Both Ways
Fylaktou asserted that Elon Musk is doing the field a service by pulling capital and public attention toward BCI.
On the other hand, he continued, “He is doing it a disservice by routinely overpromising timelines.”
“The near-term wins in this space will be quieter — motor restoration for ALS patients, closed-loop treatment for drug-resistant epilepsy and depression, interoceptive therapeutics for GI and psychiatric conditions nobody tweets about,” he contended.
“Blindsight may eventually matter,” he added, “but the field is bigger, slower, and more medically important than any single product announcement makes it look.”
