the eu recently introduced a long-anticipated set of guidelines for general-purpose ai systems, such as google’s gemini, as part of a broader push for responsible ai development. this voluntary code precedes the enforcement of the eu’s landmark ai act, which will come into effect on august 2 and aims to regulate high-risk ai technologies.
kent walker, google’s president of global affairs, confirmed the move, stating, “we will join several other companies, including us model providers, in signing the eu’s code.” yet he voiced reservations, warning that parts of the proposed framework—particularly deviations from existing eu copyright rules and rigid approval processes—could limit innovation, risk disclosing trade secrets, and slow the rollout of ai models across europe.
despite these concerns, both openai and mistral have pledged to follow the eu's recommendations. meta, on the other hand, has opted out.
earlier this month, joel kaplan, meta’s global affairs chief, criticized the framework on linkedin, arguing that it creates legal ambiguity and imposes demands that go beyond the scope of the ai act.
this standoff marks yet another point of contention between meta and eu regulators, who have previously clashed over digital rules, including those concerning political advertising. in response to such policies, meta announced it would stop accepting political ads in the eu rather than comply.
at the same time, major european corporations like airbus and lufthansa have urged the european commission to postpone implementation of the ai rules, warning that excessive regulation could hinder the continent’s competitiveness in the global ai race.