On 2 August 2025, the EU AI Act’s oblig­a­tions for gen­er­al-pur­pose AI mod­els took effect across the Union. From that date, providers plac­ing new mod­els on the mar­ket must com­ply imme­di­ate­ly with trans­paren­cy (e.g., train­ing-data doc­u­men­ta­tion) and copy­right require­ments; mod­els on the mar­ket before August 2, 2025, need to meet the same by 2 August 2027. Providers of advanced, sys­temic-risk mod­els (exceed­ing 10^25 FLOP) face addi­tion­al duties—namely, Com­mis­sion noti­fi­ca­tion and strin­gent safe­ty and secu­ri­ty measures. 

This move updates the EU’s dig­i­tal strat­e­gy to ensure clear­er infor­ma­tion on AI train­ing process­es, bet­ter enforce­ment of intel­lec­tu­al-prop­er­ty pro­tec­tions, and a secu­ri­ty-by-design approach, there­by enhanc­ing account­abil­i­ty and trust in AI across mem­ber states.

Source: Euro­pean Commission 

Accom­pa­ny­ing this sub­ject we rec­om­mend the fol­low­ing con­tent on our website