Brain-to-Brain Communication Is No Longer Science Fiction — Here's Where It Stands
From Tetris-playing brain networks to Neuralink's 21 implanted participants, the slow, strange science of linking human minds is moving faster than most people realize.
Picture two people sitting in separate rooms, miles apart, with no phones, no keyboards, no shared language. One thinks a word. The other receives it — not as a sound or an image, but as a flicker of sensation delivered directly to their visual cortex by a precisely aimed magnetic pulse. This happened in a lab. In 2014. And almost nobody talked about it for a decade.
That’s the story of brain-to-brain communication in a nutshell: astonishing progress, baffling obscurity, and a media cycle that either ignores it entirely or hyperventilates about telepathic cyborgs. Neither response is useful. What’s actually happening is weirder, more interesting, and more consequential than either camp suggests — and 2025 is shaping up to be a genuinely significant year for the field.
How this whole thing actually works
Let’s be precise, because the terminology gets slippery fast. “Brain-to-brain communication” doesn’t mean two people share a mystical mental connection. Technically, what researchers have built is a chain of technology: a brain-computer interface (BCI) reads neural signals from a sender, converts them into data, transmits that data, and then a computer-brain interface (CBI) delivers it to a receiver’s brain through some form of stimulation. The human brain isn’t directly touching another human brain. There’s a lot of silicon in between.
The signal chain looks roughly like this:
Sender wears an EEG cap or has electrodes in their motor cortex
Their brain signals are decoded in real time by an algorithm
The decoded “message” (typically a binary yes/no decision) transmits over the internet
A transcranial magnetic stimulation (TMS) device fires at the receiver’s occipital cortex
The receiver perceives a phosphene — a flash of light — that carries the encoded information
It’s elegant in the way a very long game of telephone is elegant. And the accuracy, so far, is modest. Five groups of three subjects tested the BrainNet system — the first multi-person non-invasive brain-to-brain interface — and achieved an average task accuracy of about 81%. That’s impressive for a first demonstration. It’s also a far cry from the kind of rich, nuanced communication you’d want before, say, putting this technology in a hospital.
What makes BrainNet notable is the scale. Developed at the University of Washington, the system used EEG to record brain signals and TMS to deliver information, letting three subjects collaborate on a Tetris-like task through direct brain-to-brain communication alone. Two senders decided whether to rotate a falling block; the receiver, who couldn’t see the screen, received their decisions directly via brain stimulation and acted on them. It’s Tetris. But it’s also the first working multi-person neural social network, and the jump from three brains to thirty is probably less distance than it looks. 🧠
Where Neuralink fits — and where it doesn’t
Here’s a thing that trips people up: Neuralink is not currently building a brain-to-brain communication system. What it’s building — and what it’s getting remarkably good at — is the crucial first half of that chain: reading the brain with unprecedented precision. ⚡
In February 2026, Neuralink announced 21 participants enrolled in its Telepathy program, up from the single subject who made headlines in 2024. One early user named Noland uses the implant to study languages, solve math problems, and write by controlling computers entirely through neural signals. Another participant, Brad — living with ALS and unable to speak — uses the implant to communicate outdoors, something impossible with conventional eye-tracking systems that require controlled lighting.
That’s not telepathy. But it’s a necessary precursor to it. Here’s why: any future brain-to-brain system needs two things to work at scale.
High-quality signal reading on the sender side (Neuralink’s current focus)
Precise, comfortable brain stimulation on the receiver side (still a major bottleneck)
Neuralink has made the first part dramatically better. Its device contains electrodes so thin and fragile they must be stitched into the brain by a specialized robot, and the first participant, 29-year-old quadriplegic Noland Arbaugh, reported being able to control a computer cursor and play games using only his thoughts, calling the device “life-changing.”
The company’s own long-term vision, which Elon Musk has described as “consensual telepathy,” involves sharing information between brains directly, enabling instantaneous, wordless communication — with a future “telepathic internet” that could make keyboards and even speech feel outdated. I think that timeline is very optimistic. But the direction is real. 🔬
Meanwhile, Synchron — the company backed by Bill Gates and Jeff Bezos — is taking a less invasive route. Synchron’s device doesn’t penetrate the cortex at all; founder Tom Oxley has argued that “the brain doesn’t really like having needles put into it,” and the company’s stent-based approach has already been implanted in humans. Less signal resolution, but far less surgical risk. The BCI field isn’t converging on a single design. It’s running multiple experiments in parallel, which is probably the right call when nobody is sure what the right answer looks like yet.
The gap between “binary flash” and “sharing a memory”
Here’s where I want to pump the brakes a little, because the gap between current science and science-fiction brain-to-brain communication is enormous — and worth being honest about. 🧬
The most sophisticated demonstrations to date involve transmitting a single bit of information. One flash or no flash. Rotate or don’t. Yes or no. That’s a long, long way from what most people imagine when they hear “brain-to-brain communication.” Sharing a memory, transmitting an emotion, sending a mental image — these require a completely different level of signal fidelity that current technology simply doesn’t have.
The core bottlenecks are:
Bandwidth: EEG reads from tens or hundreds of neurons at once; the human brain has roughly 86 billion. You’re getting a very blurry picture.
Decoding: Even with perfect signal capture, we still don’t know enough about how the brain encodes complex thoughts to reliably reconstruct them.
Stimulation precision: TMS can trigger a flash of light, but it can’t currently write a sentence into your visual cortex.
Individual variation: Brain activity patterns differ significantly between people, which makes building universal decoders genuinely hard.
Even recent advances in speech decoding — where researchers using implanted sensors achieved accuracy as high as 97.5% in decoding attempted speech in people with ALS — involved decoding motor patterns for speech production, not abstract thought. That distinction matters enormously. Decoding the brain’s instructions to your vocal cords is a much simpler problem than decoding what you’re thinking about. Anyone who tells you we’re close to the latter is either confused or selling something.
What I find genuinely exciting about this moment, though, is how fast speech decoding has improved. If we go from blinking Morse code to 62 words per minute via neural implants (which is where ALS patients are now, per MIT Technology Review), the next decade of compound progress starts to look interesting in a different way. What’s your intuition — is the bottleneck hardware, or is it our fundamental understanding of neural coding?
The ethics conversation nobody is having seriously enough
This is where I’ll be direct: the governance of this technology is moving nowhere near as fast as the technology itself. And that’s a problem we’ll regret. 🌍
In November 2025, UNESCO adopted the Recommendation on the Ethics of Neurotechnology, articulating a normative framework around brain-computer interfaces that emphasizes human dignity, freedom of thought, mental privacy, and autonomy. It calls on states to adopt measures preventing harmful uses, including coercive control, unlawful surveillance, or manipulation. This is real progress. A UNESCO Recommendation is also non-binding, which means it has roughly the legal force of a strongly-worded suggestion.
Chile went further: it became the first country in the world to amend its constitution to explicitly protect “neurorights,” enshrining mental privacy and integrity as fundamental rights. It’s a remarkable move, and also a reminder that every other country has not done this.
The specific concerns with brain-to-brain interfaces go beyond standard BCI data privacy worries. Oxford University philosophers Hazem Zohny and Julian Savulescu identify what might be the most profound issue: linking two minds raises entirely new questions about autonomy, identity, and the delineation of patient interests — challenges that amplify existing healthcare ethics tensions without established frameworks to address them.
Think about what that means concretely:
If information is transmitted directly to your brain and you act on it, was that action yours?
If a brain-to-brain link is hacked, what exactly gets stolen?
If two people share neural states regularly, where does one person’s identity end and another’s begin?
Because brain-to-brain transmissions can travel over the internet, one could theoretically “hack” another’s neural device — researchers have already demonstrated this kind of attack on heart pacemakers, and the principle applies. That’s not science fiction. That’s a security architecture problem waiting for regulators to notice it exists.
Nita Farahany at Duke University, author of The Battle for Your Brain, has argued that mental privacy needs to be treated as a fundamental right now, before the technology reaches scale. I think she’s right. Waiting until brain-to-brain links are commercially available to figure out the legal framework is like drafting seatbelt laws after the highways are full. The time to think hard about this is while the technology is still clunky and limited — which is exactly where we are.
What comes next, realistically
The realistic near-term trajectory for brain-to-brain communication is probably not dramatic public demonstrations. It’s incremental work happening in a few adjacent directions simultaneously. 🚀
Academic labs are continuing to refine non-invasive EEG-to-TMS pipelines, trying to push accuracy and bandwidth higher without requiring surgery. Neuralink and Synchron are racing to improve invasive BCIs primarily for therapeutic use, but the signal-reading infrastructure they’re building is dual-use: better at reading brains for medical purposes, but also potentially applicable to sending information between brains later. AI is starting to play a major role in decoding, since the pattern-matching that large neural networks excel at turns out to be very useful for interpreting the noisy, idiosyncratic signals of individual brains.
Neuralink is actively working on closed-loop systems where the brain not only sends commands but also receives feedback — mimicking natural bidirectional communication. The company also plans to expand clinical trials beyond the U.S. to Canada, the UK, Germany, and the UAE. Closed-loop bidirectional systems are the actual technical prerequisite for anything resembling real brain-to-brain communication at scale. Right now, most systems only go one direction.
The field also has serious competition from non-invasive approaches. Wearable EEG is getting better every year. Functional near-infrared spectroscopy (fNIRS) is emerging as a complementary tool. And AI decoding is dramatically lowering the bar for how clean a signal needs to be to extract useful information from it. You might not need to drill into a skull to build a functional brain-to-brain interface in 2035. That’s a legitimately open question.
If you’ve made it this far and you’re working in neurotech, BCI research, or even just thinking seriously about cognitive futures — what’s your honest assessment of the decade timeline? The binary-flash-to-shared-experience gap is the central unsolved problem. Where does the breakthrough come from?
The short, unsatisfying answer is: we don’t know. But for the first time in history, we have 21 people walking around with working brain implants, researchers who’ve demonstrated multi-brain collaborative networks, a budding international regulatory conversation, and enough published science to stop treating this as speculation. Brain-to-brain communication is real, limited, and accelerating. The question isn’t if anymore. It’s how fast — and who gets to decide how it’s used.