Meta has confirmed it will not sign the European Union's voluntary code of conduct for general-purpose artificial intelligence, just two weeks before the bloc's binding Artificial Intelligence Act takes effect on August 2, 2025. The social media giant cited "legal uncertainty" introduced by the voluntary guidelines and argued they exceed the scope of existing regulations, creating potential compliance risks for companies.

EU officials, however, have defended the code, emphasizing that it was designed to ease adherence to the upcoming mandatory rules. They note the voluntary framework offers clear benchmarks to demonstrate compliance, helping businesses avoid penalties under the stricter legislation. Violations of the Artificial Intelligence Act could result in fines of up to 7% of a company's global revenue—a significant stakes for Meta, whose annual earnings run into tens of billions of dollars.

The standoff highlights growing tensions between tech firms and European regulators over AI governance. The voluntary code, launched in 2023, was intended as a transitional measure to encourage responsible AI development ahead of the Act's enforcement. It outlines best practices for transparency, bias mitigation, and risk management in AI systems.

Meta's refusal underscores concerns within the industry that voluntary commitments might set precedents conflicting with the final legal text of the Artificial Intelligence Act. A company spokesperson stated, "We support robust AI regulation but cannot endorse guidelines that diverge from the statutory framework, as this would create confusion for both businesses and users."

EU digital policy chief Thierry Breton countered that the code "aligns closely with the Act's objectives" and accused Meta of missing an opportunity to build trust. "Voluntary compliance now paves the way for seamless adaptation to mandatory rules," he said in a statement.

As the August 2 deadline approaches, the dispute raises questions about how other major AI developers will navigate the transition. While several tech companies have signed the code, Meta's reluctance signals potential challenges in harmonizing industry practices with Europe's landmark AI regulations.