Provisions of the European Commission’s (EC) AI Act that affect general-purpose AIs are set to go into effect this August. But leaders of both US and EU companies, and even some EU heads of state, have called the legislation premature and are asking for a pause.
The AI Act first went into effect this February, banning AIs that fall into its “unacceptable risk” category, such as social scoring systems and AIs that scrape the internet or CCTV for facial recognition data.
On August 2, “general-purpose” AIs, or those with wide uses that can be integrated into many different applications, will be required to comply with EU copyright law, disclose their technical documents, and provide the EC with a summary of their training data. The AI Act will be fully in effect, minus a few exceptions, by August 2026, according to the EC’s website.
Underbaked regulation? But some companies and politicians have argued that much of the Act isn’t ready for prime time. As the Center for European Policy Analysis (CEPA) pointed out, some of the AI standards are still being drafted, and some EU member states have yet to establish AI oversight bodies.
In an open letter to the EC, 46 leaders of large EU companies—including Airbus, Lufthansa, microchip lithography company ASML, and French AI startup Mistral—asked for a “two-year ‘clock-stop’” on the AI Act. Swedish prime minister Ulf Kristersson and other EU leaders have requested a pause as well.
News built for finance pros
CFO Brew helps finance pros navigate their roles with insights into risk management, compliance, and strategy through our newsletter, virtual events, and digital guides.
So have some US companies, including Google and Meta. The Act will affect many US companies that have clients or customers in the EU. Businesses with direct operations in the EU, whose AI might touch EU companies in their supply chains, or whose AI models process EU citizens’ data, should see whether the Act applies to them, KPMG recommends.
Uncertainty abounds: The EC has stated that the AI Act will proceed as scheduled. “Let me be as clear as possible, there is no stop the clock. There is no grace period. There is no pause,” EC spokesperson Thomas Regnier said at a press conference, Reuters reported.
But Henna Virkkunen, the EC’s tech head, has said she’ll decide by the end of August whether a pause is needed. CEPA predicts that the February ban on unacceptably risky AI models will stay in place, but that regulatory obligations for high-risk and general-purpose AIs will be pushed out one to two years.
CFOs, along with their CIO and CISO colleagues, will definitely want to watch this space.