Are we moving towards global regulation of AI?

There are a num­ber of issues to con­sider. First, there is the fear of an exist­en­tial risk fuelled by stor­ies of an AI that could become autonom­ous and des­troy human­ity. How­ever, I don’t see this as a real pos­sib­il­ity – at least not with the mod­els that are being developed right now. Then there’s the fear of an oli­go­poly, with heavy depend­ence on a hand­ful of com­pan­ies. There is also the risk of human rights viol­a­tions. And in this case, a new factor comes into play: for a long time, the most ter­rible actions were reserved for the coun­tries and armies of the most power­ful coun­tries. Now it’s com­mon­place, mass-mar­ket tech­no­logy. The same AI that can detect can­cer could also be used to pre­vent part of the pop­u­la­tion from enter­ing air­ports. We have rarely seen such multi-pur­pose tech­no­lo­gies. AI also makes mali­cious acts pos­sible, such as deep fakes, attacks on our sys­tems, or the manip­u­la­tion of opin­ion. Not to men­tion the imbal­ances it will cre­ate in terms of intel­lec­tu­al prop­erty, pro­tec­tion of the pub­lic domain, pro­tec­tion of pri­vacy, and the upheavals in the work­place. All these chal­lenges, coupled with the desire to really bene­fit from AI’s bound­less prom­ise, have promp­ted intense glob­al reflec­tion on a frame­work for shared governance.

The greatest risk, in my opin­ion, is that of a mono­poly. It’s a ques­tion of demo­cracy. One of the great dangers is that the glob­al eco­nomy, soci­ety and the media will become ultra-depend­ent on a small oli­go­poly that we are unable to reg­u­late. In line with this ana­lys­is, I’m try­ing to push for the pro­tec­tion of the digit­al com­mons and open source. We need to make sure that there are mod­els that are free to use so that every­one can bene­fit from them. There are already high-qual­ity open source mod­els. So, the ques­tion is: are we going to put enough pub­lic money into train­ing them for the bene­fit of every­one? In my opin­ion, that’s where the real battle lies. Ensur­ing that there are resources for the pub­lic good and that it will be pos­sible to innov­ate without ask­ing for permission.

Among the aspects that could lead to inter­na­tion­al coordin­a­tion are the indir­ect effects of AI. These tech­no­lo­gies could dis­rupt the way in which we have con­struc­ted the pro­tec­tion of pri­vacy. Today, the gen­er­al prin­ciple is to pro­tect per­son­al data in order to pro­tect indi­vidu­als. With AI, by using pre­dict­ive mod­els, we can learn a great deal, if not everything, about a per­son. Pre­dict­ive mod­els can take into account a person’s age, where they live and where they work, and give a very good prob­ab­il­ity of their risk of can­cer or their like­li­hood of lik­ing a par­tic­u­lar film. Anoth­er issue being debated at an inter­na­tion­al level is that of intel­lec­tu­al prop­erty. It is now pos­sible to ask an AI to pro­duce paint­ings in the style of Keith Har­ing and to sell them, which poses a prob­lem for the artist’s beneficiaries. 

Ten years ago, there was a tacit agree­ment not to reg­u­late social net­works. Now these com­pan­ies are worth $1,000bn and it’s very dif­fi­cult to change their tra­ject­ory. Most developed coun­tries are telling them­selves that they won’t make the same mis­take again and that they mustn’t miss the boat this time. But that means they know what to do. There is a new aware­ness that reg­u­la­tion must be anchored in an inter­na­tion­al frame­work. There are so few bor­ders in the digit­al world that you can set up in any coun­try and still have a pres­ence in anoth­er. Clearly, there is intense inter­na­tion­al activ­ity around the major issues men­tioned above: exist­en­tial risk, the ques­tion of eco­nom­ic sov­er­eignty and mali­cious actors.

No. The “region­al” level (i.e. a coher­ent group of coun­tries) is also very import­ant. Europe has learned the hard way that only reg­u­lat­ing all digit­al uses at nation­al level is not enough to bring the giants to heel. When we estab­lish a European frame­work, they nego­ti­ate. But that cre­ates oth­er ten­sions, and we don’t want to encour­age an inter­na­tion­al order based on extra-ter­rit­ori­al applic­a­tion decisions. So, the idea that the inter­na­tion­al level is the right size for think­ing about the digit­al world has taken hold, and it’s not really ques­tioned any more.

Under the law, we have the right to pro­hib­it the devel­op­ment of cer­tain AI. But we are afraid that the oth­er powers will con­tin­ue to do so, and that we will become weak and obsol­ete. It’s a ques­tion of being both pro-innov­a­tion and pro-secur­ity for cit­izens, and that’s why every­one would like decisions to be col­lect­ive. These tech­no­lo­gies are chan­ging very fast, and cre­at­ing a lot of power, so we don’t want uni­lat­er­al disarmament.

The eth­ic­al frame­work is being dis­cussed. Dis­cus­sions are tak­ing place in dozens of for­ums with­in busi­ness, civil soci­ety, among research­ers, at the UN, the G7, the OECD and as part of the French ini­ti­at­ive for a glob­al part­ner­ship for arti­fi­cial intel­li­gence. It’s also a dip­lo­mat­ic effort that’s com­ing out of the embassies, with debates at the Inter­net Gov­ernance For­um and the annu­al sum­mit on human rights, RightsCon. Little by little, ideas are tak­ing root or becom­ing estab­lished. We are still in the pro­cess of identi­fy­ing the con­cepts on which an agree­ment will be based. An ini­tial con­sensus is emer­ging around cer­tain prin­ciples: no use should be made that is con­trary to human rights; the tech­no­logy must be in the interest of the users; it must be proven that pre­cau­tions have been taken to ensure that there is no bias in the edu­ca­tion of the mod­els; the need for trans­par­ency so that experts can audit the mod­els. Then it will be time to look for treaties.

There are also debates about a demo­crat­ic frame­work. How can we ensure that these com­pan­ies are not manip­u­lat­ing us? Do we have the right to know what data the AI has been trained with? The notions of secur­ity in the face of exist­en­tial risk were much dis­cussed in the UK at last year’s world sum­mit. Now the con­ver­sa­tion is turn­ing to the future of work and intel­lec­tu­al prop­erty, for example. In 2025, France will be host­ing a major inter­na­tion­al sum­mit on AI, which will help move these issues forward.

Many people think not. Per­son­ally, I think that a good text is fun­da­ment­al. The Declar­a­tion of Human Rights is still val­id, but tech­no­lo­gies have changed. The 1978 Data Pro­tec­tion Act has become the GDPR, but the prin­ciple of user con­sent for data to be cir­cu­lated has not aged a day. If we can find robust prin­ciples, we can pro­duce texts that will stand the test of time. I think we could reg­u­late AI with the GDPR, the respons­ib­il­ity of the media and con­tent pub­lish­ers, and two or three oth­er texts that already exist. It’s not a giv­en that we need an entirely new framework.

Sirine Azououai