European Commission launches the AI Pact ahead of AI Act deadlines

The European Commission launched the AI Pact, a voluntary scheme inviting companies to commit early to obligations under the AI Act before those obligations became legally binding, convening industry stakeholders in Brussels to mark the first signatories. The Act itself had entered into force on 1 August 2024, with its provisions phasing in over the following years rather than applying immediately.

Signatories committed to three core actions: adopting an internal AI governance strategy, identifying and mapping which of their AI systems would count as high-risk under the Act’s classification, and promoting AI awareness among their own staff. More than half of the initial signatories went further, making additional pledges on human oversight, risk mitigation or transparency around AI-generated content such as deepfakes. Over a hundred companies signed on at launch, spanning IT, telecommunications, healthcare, banking, automotive and aeronautics, and ranging from large multinationals to small and medium enterprises across the EU and beyond; the Commission’s announcement did not name individual signatories.

The pledges carried no legal force; the Commission described the Pact explicitly as preparation rather than compliance, a bridge for firms to get ahead of enforcement powers and high-risk-system requirements that would not become applicable for years afterward. As with similar voluntary frameworks elsewhere — the White House’s earlier voluntary AI commitments among them — the Pact’s practical significance lay less in what it obliged signatories to do than in establishing, ahead of binding law, which companies were willing to be publicly associated with the compliance effort at all.