General-Purpose AI Code of Practice (2025)
A voluntary code published by the European Commission in July 2025 that lets providers of general-purpose AI models demonstrate compliance with the European regulation.
The European AI regulation imposes on providers of general-purpose models obligations of transparency, copyright compliance and, for models with systemic risk, risk assessment and mitigation; it does not say how to meet them. The code translates those obligations into concrete commitments, in three chapters: transparency, copyright, safety and security. Signing is not mandatory, but it gives a recognized path to compliance, and a provider that departs from it must prove compliance some other way. For the insurer of a provider or an integrator, the signature and the documentation it produces become underwriting inputs.
The Commission published the code on July 10, 2025, three weeks before the obligations of general-purpose model providers took effect on August 2, 2025; most large providers signed it, while one major US provider publicly announced its refusal.
Follow the research
One letter a week. First, the week's underwriting note, when one is published: a dated, argued decision, signed by Paul Bernardski. Then what the observatory published on cyber risk, political risk and insurance. No account, no tracking.
No spam, one-click unsubscribe, no data ever sold.