Brain implants: the true, the false and the uncertain

A brain-machine inter­face that can be implanted in a single morn­ing: that is the prom­ise of Neur­alink¹, a start-up co-foun­ded by Elon Musk, which tested its first brain implant on a patient in 2024. The goal is simple, to restore lost func­tions to para­lysed patients and, ulti­mately, to enhance human phys­ic­al and men­tal capabilities.

Tech giants are ready to invest bil­lions of dol­lars in neur­o­tech­no­lo­gies, par­tic­u­larly in the devel­op­ment of neur­al implants. Inser­ted into the brain, these tech­no­lo­gies raise tech­nic­al, con­cep­tu­al and eth­ic­al ques­tions. Clé­ment Hébert, spe­cial­ist in neur­al implants at the Gren­oble Insti­tute of Neur­os­cience, and Her­vé Chnei­weiss, a neur­os­cience research­er and chair of the INSERM eth­ics com­mit­tee, answer our ques­tions about these unique pros­thet­ic devices.

FALSE

Clé­ment Hébert. The first brain pros­thet­ics ini­tially tar­geted sens­ory func­tions. The very first coch­lear implant to restore hear­ing in deaf people was developed in 1961, before being mar­keted in the 1980s. This elec­tric­al brain-machine inter­face con­verts sound into elec­tric­al sig­nals for the brain, effect­ively repla­cing the coch­lea, which is no longer able to restore hear­ing. Then, in the 1990s, per­man­ent implants were installed in patients with Par­kin­son’s dis­ease to reduce tremors.

Since the 2000s, there has been a tech­no­lo­gic­al break­through with the use of micro­elec­tron­ics to cre­ate increas­ingly smal­ler implants that can be more eas­ily integ­rated into the body. Pros­thet­ics now enable research­ers to record the activ­ity of sev­er­al thou­sand neur­ons using micro­elec­trode arrays in very spe­cif­ic areas of the brain and, in turn, stim­u­late it to per­form spe­cif­ic func­tions. For example, con­trolling a curs­or on a screen using an implant was explored at Brown Uni­ver­sity in the United States in the 2010s.

In 2023, the Swiss Fed­er­al Insti­tute of Tech­no­logy in Lausanne and the Clin­atec Centre at the CEA Gren­oble helped a para­ple­gic patient walk again using a brain implant and an elec­tric­al stim­u­la­tion device in the spin­al cord. In 2025, a research team at the Uni­ver­sity of Cali­for­nia restored speech to para­lysed patients using implants cap­able of decod­ing their inten­tion to speak and tran­scrib­ing it via a speech syn­thes­iser. Neur­alink fol­lowed suit in 2024, build­ing on exist­ing tech­no­lo­gic­al advances but with optim­ised elec­tron­ics, enabling it to send more detailed information.

UNCERTAIN

Her­vé Chnei­weiss. Implants were ori­gin­ally used to restore lost func­tions, but it is con­ceiv­able that exist­ing func­tions could be built on in a tar­geted and lim­ited way. For example, labor­at­ory tri­als show that cer­tain com­pon­ents of memory can be enhanced. By admin­is­ter­ing elec­tric shocks to the hip­po­cam­pus, we find that memor­ies formed imme­di­ately after the small shock are bet­ter retained. Apart from that, transhuman­ist ideas that humans will become more intel­li­gent thanks to implants are pure fantasy!

FALSE

CH. The dream of increas­ing com­put­ing power of the human brain by boost­ing it is not bio­lo­gic­ally feas­ible. The com­put­ing speed of a brain is in the order of mil­li­seconds, where­as a com­puter oper­ates in nano­seconds, or even faster. In oth­er words, the human brain cal­cu­lates a thou­sand times slower than a com­puter and there­fore can­not be as effi­cient as a com­puter in pro­cessing inform­a­tion, even if it is sent massive amounts of new inform­a­tion or stim­u­la­tion. For now, the tech­no­lo­gies developed con­sist of an intern­al part in the brain (implant) con­nec­ted to an extern­al part (com­puter). The implant con­sists of a net­work of elec­trodes that are in con­tact with the neur­al tis­sue and a com­mu­nic­a­tion sys­tem that sends sig­nals to the elec­trodes or receives sig­nals col­lec­ted by the electrodes.

Extern­ally, com­puter sys­tems can record, pro­cess and decode data and send inform­a­tion to one or more effect­ors, such as a robot­ic arm, for example, to restore motor func­tion. Ulti­mately, what we will be able to achieve is to inter­n­al­ise a com­puter, but we will still need to con­tin­ue to query it. For me, the only thing we will be able to increase is the speed of com­mu­nic­a­tion with a com­put­ing sys­tem! And in that case, we may ask ourselves who is doing the com­put­ing? Me, as a human being, or the micro­chip inside me?

UNCERTAIN

HC. In medi­cine with­in the European Uni­on neur­al data is con­sidered ‘sens­it­ive per­son­al data’ and is there­fore pro­tec­ted by the rules of the Gen­er­al Data Pro­tec­tion Reg­u­la­tion (Art­icle 9 of the GDPR). Inter­na­tion­ally, bod­ies that are con­sid­er­ing these issues, such as the Organ­isa­tion for Eco­nom­ic Co-oper­a­tion and Devel­op­ment (OECD) and the United Nations Edu­ca­tion­al, Sci­entif­ic and Cul­tur­al Organ­isa­tion (UNESCO), want neur­al data, even out­side the med­ic­al con­text, to be con­sidered per­son­al and sens­it­ive².

To pro­tect human rights in the use of pro­cesses that enable the inter­pret­a­tion of brain activ­ity for med­ic­al pur­poses or com­mer­cial applic­a­tions, a set of recom­mend­a­tions con­cern­ing the eth­ics of neur­o­tech­no­lo­gies was adop­ted by UNESCO mem­bers in Novem­ber 2025 (before the United States with­drew). The field of neur­al implants is covered in these recom­mend­a­tions. It was decided to write non-bind­ing recom­mend­a­tions so that the text could be adop­ted by as many states as pos­sible. Thus, the UNESCO text, which con­tains more than 160 recom­mend­a­tions, became the first ref­er­ence text for all coun­tries around the world. How­ever, each state will be able to take it and inter­pret it accord­ing to its own laws.

TRUE

CH. Cur­rently, there are only about fifty patients world­wide who have port­able implants such as those developed by Clin­atec or Neur­alink. And there is a good reas­on for that. These brain-machine inter­faces are still far from being truly oper­a­tion­al, even in clin­ic­al set­tings. The cur­rent goal is to devel­op effect­ive implants that remain oper­a­tion­al for dec­ades, because, for the moment, research into these neur­o­tech­no­lo­gies faces sev­er­al chal­lenges. For example, find­ing mater­i­als that are thin and flex­ible enough not to cause too much trauma or rejec­tion reac­tions. In addi­tion, cur­rent flex­ible tech­no­lo­gies are sens­it­ive to water over the long term, affect­ing the lifespan of implants. Finally, exist­ing tech­no­lo­gies for trans­mit­ting and pro­cessing neur­al sig­nals (so that inform­a­tion can be trans­mit­ted wire­lessly between the implant and the com­puter) cause a loc­al rise in tem­per­at­ure that is harm­ful to neur­al tissue.

HC. The wide­spread use of brain-machine inter­faces would fall under the umbrella of what is known as neuro­cos­met­ics, the equi­val­ent of cos­met­ic sur­gery for the brain. In Europe, this will con­tin­ue to be con­sidered a med­ic­al pro­ced­ure and will there­fore be sub­ject to strict reg­u­la­tions. It is there­fore unlikely that European reg­u­lat­ors will author­ise the implant­a­tion of chips in the brain without med­ic­al neces­sity. In the United States, Neur­alink was able to install its first implants because the US Food and Drug Admin­is­tra­tion (FDA) author­ised it to do so in a med­ic­al set­ting for patients with a med­ic­al need.

More gen­er­ally, there is a purely con­cep­tu­al ques­tion sur­round­ing this type of invas­ive tech­no­logy: are we pre­pared to have a chip implanted (even on the sur­face) in our brains to play video games or con­trol our phones? I believe that the future of neur­o­tech­no­logy lies in non-invas­ive tech­no­logy. For example, Google has developed earphones with elec­tro­en­ceph­al­o­gram (EEG) elec­trodes, and aca­dem­ic labor­at­or­ies such as the Mas­sachu­setts Insti­tute of Tech­no­logy’s Media Lab are work­ing on EEG glasses that can be worn on the arms to detect ten­sion and men­tal fatigue or move a cursor.

Interview by Olympe Delmas