Europe must do more to harness its AI ambitions

Financial Times Europe6 Aug 2026The writer is a fel low at Stan ford Uni versity’s Insti tute for Human Centered Arti fi cial Intel li gence and author of ‘The Tech Coup’Mari etje Schaake

European Coun cil pres id ent António Costa is schedul ing a dis cus sion amongEuropean heads of state about AI this autumn. Bet ter late than never. Lead ers mustuse this oppor tun ity to catch up and align on AI policy. While the US and China treatthe tech no logy as a key pri or ity and instru ment of state power, the EU has not evendecided on a col lect ive polit ical course.

Today, any dis cus sion of digital sov er eignty or stra tegic autonomy is, in effect, a dis -cus sion about AI. This global gamble cre ates a risk for Europe. Without sov er eign AIoptions, Europeans are depend ent on for eign plat forms and tools. These depend en -cies go bey ond com mer cial decisions. They mean that decisions made by for eigngov ern ments have a dis pro por tion ate impact on European national secur ity.

Polit ical lead ers must there fore con front the AI side of secur ity head-on as ten sionbuilds between AI labs and geo pol it ics. They should pri or it ise trust and safety.

In the US, where invest ment in AI is break ing all records, the Trump admin is tra tionregards AI as a source of its hege monic global ambi tions. Pres id ent Don ald Trumpis tak ing ever more steps to boost US advant ages in AI capa city, access and agency.Reports sug ges ted that Anthropic’s latest Mythos AI model was being used by USintel li gence agen cies, while access for European allies was lim ited.

This presents a new secur ity chal lenge for the EU. The recent Nato sum mit in Ank -ara saw European lead ers con front the real ity of an unre li able and pos siblyadversarial Amer ica that forces the EU to develop and main tain its own mil it aryand defence capa cit ies. Cyber se cur ity is integ ral to resi li ence. Tech no lo gical devel -op ments also make it clear that divor cing eco nomic choices from secur ity object -ives is impossible. Cor por ate actions have an increas ing impact on safety and secur -ity.

AI is at the heart of this incon veni ent truth. To tackle poten tial threats from cyberincid ents or exploit a tions of depend en cies, EU states must set up struc tures to co-ordin ate emer gency responses and bene fit from shar ing les sons learnt. The sum mitshould decide on those.Ded ic ated task forces have cre ated urgency and focus dur ing past crises in Europe,includ ing Covid and the fin an cial crisis. They are mer ited here too. A clear AI policypack age should include hir ing AI tal ent into policy pos i tions and devel op ingnational AI secur ity insti tutes, as well as a broader tal ent pro gramme to grow andattract AI tal ent (back) to Europe.

Within the con tin ent there already exists tre mend ous energy and appet ite forstronger EU lead er ship on AI. National AI strategies have been announced andadop ted, some uni versit ies are decoup ling from non-EU tech pro viders and intel li -gence agen cies are choos ing domestic sys tems over those offered by inter na tionalcom pan ies. Tying such bot tom-up decisions with an EU-wide vis ion should scaleand boost the bloc’s com pet it ive pos i tion by divid ing labour between mem ber statesand lever aging cap ital incent ives, ideally in sync with industry. European lead erscan do a real ser vice to people world wide who are deeply unhappy with the two fla -vour menu of AI they are cur rently offered.

Being ambi tious polit ic ally does not mean pri or it ising AI above everything else. Yes,Europe can choose to do AI dif fer ently and not copy the extract ive Sil icon Val leyway of redu cing people to con sumers and soci et ies to mar kets. Europe is and willremain far removed from the Chinese model of state-dir ec ted com pan ies andinnov a tion. But being prin cipled and right on paper trans lates into very few res ultsin today’s data, cap ital and com pute driven world. If Europeans believe AI can addvalue while respect ing demo cratic val ues and increas ing safety, they must show so.

The sum mit this autumn must have a clear agenda and real ambi tion from whichEuropean lead ers walk away with a com pre hens ive AI strategy.

It should there fore con clude with ser i ous polit ical buy-in and a tan gible strategythat unites the lever age that comes from Europe’s single mar ket and wealth:incentiv ising invest ment into sov er eign capa city such as com pute, mak ing AIchoke points more resi li ent, and set ting pri or it ies for which capa cit ies must benation ally held first.

Any strategy is only as good as its exe cu tion. The sum mit must provide con clu sionsand actions. The EU can not afford to waste this oppor tun ity and spend the autumnmerely learn ing and brain storm ing about AI