Your Thoughts Are the Next Data Goldmine. Nobody Is Protecting Them.
Maryam Tariq
Brain chip companies are in a commercial sprint. AI has proven it can intercept your silent inner voice. And the legal framework to stop misuse does not exist anywhere on earth.
The Thought That Was Never Meant to Be Heard
In 2025, Stanford University researchers studying brain-computer interfaces made a discovery that went largely unreported in its most alarming dimension. They were not just demonstrating that paralysed patients could communicate via brain signals. They were showing that AI could detect spontaneous inner speech, words and thoughts a person never consciously tried to express, by passively monitoring electrical activity in the motor cortex.
The team designed tasks where participants silently counted shapes on a screen, expecting they would count internally. The AI picked up the number words passing through their brains without any deliberate attempt to communicate them. For structured imagined sentences, inner speech decoding reached 74% accuracy in real time. The signals were weaker than those produced by attempted speech, but they were there, readable, and extractable.
This is not a capability that exists only in a California research lab. It is a capability that several well-funded private companies are actively working to put inside a consumer product.
The Companies Competing to Own Your Neural Interface
The commercialisation race for brain-computer interface technology is already well underway. Neuralink, founded by Elon Musk, has moved from animal trials to human implantation and has stated ambitions that extend far beyond medical use into general consumer enhancement. Synchron, an Australian-American company, has developed a less invasive approach using a stent-like electrode inserted through blood vessels into the brain, already trialled in patients with ALS. Meta, meanwhile, is developing a non-invasive wrist-based neural band that intercepts nerve signals traveling from the brain to the hand, positioning it as a control interface for augmented reality devices.
Maitreyee Wairagkar, a neuroengineer at UC Davis who has been central to advancing speech BCI technology, has stated plainly that commercialisation at scale will begin within the next few years. The technology achieving 97.5% accuracy decoding speech at 32 words per minute from an ALS patient today will not stay in the clinic. The market logic pulls it outward.
Brain Data Is Unlike Any Data That Has Come Before
Every category of personal data collected today carries risk. Location data reveals your movements. Financial data reveals your choices. Search history reveals your curiosity and sometimes your fears. But brain data occupies a fundamentally different category.
Neural signals can potentially reveal what a person is about to say before they say it, what they genuinely feel as distinct from what they perform socially, whether they are experiencing fear, arousal, or deception, and the contents of thoughts they had no intention of sharing with anyone. Today’s implanted devices sample around 256 neurons simultaneously. The human brain contains approximately 86 billion neurons. Researchers at UC Davis and Stanford have both acknowledged that scaling electrode counts is the clearest path to richer, more accurate decoding. The devices of the next decade will not sample 256 neurons. They will sample thousands, then millions.
A Legal Framework That Does Not Exist
The European Union’s General Data Protection Regulation is widely considered the world’s most comprehensive personal data law. It does not specifically classify neural signals as a protected category. The United States Health Insurance Portability and Accountability Act covers medical data within healthcare contexts. It was not designed for a consumer brain chip sold by a technology company. Neither framework includes provisions for cognitive liberty, the right not to have your thoughts monitored, extracted, or used without consent.
Chile became the first country in the world to enshrine neurorights in its constitution in 2021, adding protections for mental privacy and cognitive freedom. It remains the only country to have done so. No major technology market, not the United States, not the European Union, not China, has followed with enforceable, purpose-built neural data legislation.
There are currently no legal requirements for consent before neural data is collected by a commercial BCI device. There are no data minimisation rules specifying that a brain chip may only record what is necessary for its stated function. There is no prohibition on secondary use of neural data, meaning a company that collects your brain signals to let you control a cursor is not legally barred from using those same signals to infer your emotional state, attention level, or purchasing intent.
The Window to Act Is Now
The pattern is familiar. Social media platforms deployed at scale before meaningful privacy regulation existed. The result was a decade of reactive legislation, inadequate enforcement, and a structural power imbalance between platforms and users that has never been corrected. Neurorights Foundation founder Rafael Yuste and other researchers have argued urgently that the same mistake cannot be made with brain data, because the intimacy of what neural signals reveal makes the social media era look comparatively benign.
Stanford’s Frank Willett has noted that his lab’s next research priority is exploring the superior temporal gyrus, an auditory brain region that may hold even richer inner speech representations than the motor cortex. UC Davis plans to develop devices that sample far more neurons than today’s 256 electrode arrays. Each technical step forward is also a step toward a more complete picture of a person’s inner life being accessible to an external system.
The technology is not pausing to wait for policy. The commercialisation race is active. The privacy stakes are unprecedented. And the legal framework that should govern all of it does not yet exist.
Sources:
· BBC Science Focus, Intelligence Revolution series, February 2026: https://www.bbc.com/future/article/20260227-how-ai-can-read-our-scrambled-inner-thoughts
· Stanford Neural Prosthetics Translational Laboratory, Frank Willett: https://neuralinterfaceslab.stanford.edu
· UC Davis Neuroprosthetics Laboratory, Maitreyee Wairagkar: https://npl.ucdavis.edu
· Neuralink official: https://neuralink.com
· Synchron brain-computer interface: https://synchron.com
· Neurorights Foundation, Rafael Yuste: https://neurorightsfoundation.org
· Chile neurorights constitutional amendment, 2021: https://www.bcn.cl/leychile/navegar?idNorma=1171191
· European Parliament briefing on neurotechnology and GDPR: https://www.europarl.europa.eu/thinktank/en/document/EPRS_BRI(2021)690036
