The European Union’s efforts to introduce voluntary guidelines for managing advanced artificial intelligence models have come under fire from major tech players, casting doubt on the initiative’s success ahead of the AI Action Summit in Paris.

The proposed “code of practice” aims to provide a framework for companies like OpenAI, Google, Meta, and Microsoft, outlining principles for handling general-purpose AI models such as GPT and Llama.

It follows last summer’s adoption of the EU’s Artificial Intelligence Act, offering clarity on issues like data transparency, systemic risks, and third-party testing. However, industry giants argue the draft rules are overly burdensome and risk stifling innovation.

Google’s Kent Walker, president of global affairs, called the plan a “step in the wrong direction” at a critical time for Europe’s competitiveness. Speaking to POLITICO, Walker criticised elements of the draft, including its provisions on copyright and third-party model testing, as redundant or overly demanding. He hinted that Google’s decision to sign the code could hinge on discussions at the Paris summit.

Meta’s top lobbyist, Joel Kaplan, expressed even sharper opposition, describing the proposed requirements as “unworkable and technically unfeasible.” During a livestream, Kaplan stated that Meta would not sign the code in its current form, arguing that it exceeds the scope of the AI Act.

Kaplan also criticised the EU’s broader regulatory approach, labeling its hefty fines on U.S. tech companies as “tariffs.”

He pointed to Meta’s €797 million fine over the dominance of Facebook Marketplace as an example of excessive penalties that neither benefit consumers nor foster competition. Kaplan argued that this punitive mindset undermines Europe’s competitiveness in the global AI race, warning that the EU risks missing out on the current wave of AI innovation.

Recommended

• Twitch Content Labelling Vastly Improved, Says Ofcom
• TikTok Trades Human Content Moderators for AI
• EU Users Can Settle Social Media Content Disputes Out of Court

The criticism comes at a pivotal moment, as European Commission President Ursula von der Leyen and other leaders seek to showcase the EU as a business-friendly region at the Paris summit. Work on the code of practice is expected to conclude in April, but its success largely depends on whether tech giants like Google and Meta sign up.

Meta has also announced significant changes to its content moderation strategy, phasing out its fact-checking programme in the U.S. in favor of a user-driven system modeled on X’s community notes. Meta has previously revealed plans to expand the programme globally by 2026, including eventual implementation in the EU, with assurances that the company would work closely with European regulators.