The Brain-Computer Interface Patent Race

The narrative around brain-computer interfaces treats the field as a product of the last decade. Neuralink files for FDA approval. Synchron beats it to the first US human trial. Meta announces non-invasive neural wristbands. The coverage positions these as the beginning of something. The patent record positions them as the commercial phase of something that started much earlier.

The foundational intellectual property in BCI technology was filed between the late 1960s and the early 2000s, largely by defense contractors, university research programs, and individual inventors working in military-adjacent research environments. By the time any of the currently prominent companies were founded, the core concepts had been patented, litigated, expired, and in some cases re-patented in narrower forms by subsequent inventors.

The earliest patents in the BCI space addressed the problem of detecting and interpreting brain signals using external sensors. The electroencephalogram had been a clinical tool since the 1920s. The first patents attempting to turn EEG signals into control inputs for external devices appeared in the 1960s. These were crude systems by any current standard. They identified the direction before the engineering existed to pursue it seriously.

The Malech patent of 1973 is the most significant early filing because it goes beyond detection into alteration. Most EEG-adjacent patents of that period described reading signals. Malech described a bidirectional system: read the signals, process them, transmit modified signals back. That bidirectional architecture is exactly what current advanced BCI programs are working toward. DARPA's Neural Engineering System Design program, announced in 2017, described a target of reading and writing neural signals simultaneously at the scale of one million neurons. Malech described the same architecture at the patent level in 1973, without the electrode density or wireless bandwidth that current technology provides.

The 1990s produced the academic research that the current commercial wave is built on. Philip Kennedy and Roy Bakay at Emory University implanted a neurotrophic electrode in a locked-in patient in 1996 and demonstrated cursor control through neural signals in 1998. Miguel Nicolelis at Duke University demonstrated real-time control of a robotic arm by a monkey using implanted electrode arrays in 2000. John Donoghue at Brown University developed the Utah Array, the electrode architecture that became the basis for BrainGate and several subsequent commercial programs.

Each of these research streams produced patents. Some were held by universities. Some were assigned to the research teams' spinout companies. Some were filed by researchers individually. By 2010, the patent landscape around motor-cortex BCI technology was dense enough that any new commercial entrant faced a significant freedom-to-operate analysis before filing its first product patent.

Parallel to the academic research, DARPA funded BCI programs throughout the 1990s and 2000s under program names including Revolutionizing Prosthetics, Accelerated Learning, and Cognitive Technology Threat Warning System. Some of the research these programs produced entered the public domain through academic publication. Some entered the patent system through contractor filings. Some remains in classified program records.

The Revolutionizing Prosthetics program funded the development of the DEKA arm, a prosthetic limb controlled by neural signals that reached clinical trials by 2014. The Accelerated Learning program investigated whether neural stimulation could accelerate the acquisition of complex skills. The Cognitive Technology Threat Warning System investigated whether neural signals could detect threats in surveillance imagery faster than conscious visual processing allowed. Each of these programs produced technical advances that were partially documented and partially not.

The companies currently described as competing in the BCI space are not racing to invent the foundational technology. They are racing to be the first to bring specific applications to regulatory approval and commercial deployment at scale. The underlying neuroscience and the core engineering concepts have been established. The race is about miniaturization, wireless bandwidth, electrode longevity, surgical procedure simplification, and regulatory navigation.

Neuralink's specific technical contribution, if it has one, is in the surgical robot it developed to implant its electrode arrays with precision that reduces tissue damage. Synchron's distinction is in its endovascular approach, which avoids open-brain surgery entirely by threading a stent-electrode through blood vessels. Neither company invented the concept of reading motor cortex signals and translating them into device control. Both are refining the delivery mechanism for a concept that was demonstrated academically two decades ago and patented at the foundational level fifty years ago.

Current commercial BCI programs describe their systems as read-only or primarily read-oriented. They take signals from the brain and translate them into external outputs. The Malech patent described a read-write system. DARPA's NESD program described a read-write target. The neuroscience of bidirectional neural interfaces, systems that both read neural signals and write signals back into neural tissue, is an active research area. It is not an active commercial announcement area.

Writing signals into neural tissue is technically more complex than reading them and raises regulatory and ethical questions that reading does not. It also describes a capability that, if reliable and precise, would represent something qualitatively different from any currently approved medical device. The foundational patent for that capability was filed in 1973. The research programs pursuing it at scale are partially classified. The commercial announcement has not been made.

---

The brain-computer interface patent record runs from the late 1960s through the present. The foundational concepts were established in the first decade of that record. The defense research programs that developed those concepts into working systems produced results that are partially public and partially not. The commercial companies now described as racing to build the future of neural interfaces are commercializing a technical foundation built over fifty years in research environments that were not designed to make their work visible.