The final Carolina Principles joint statement calls for adapting regulation to genuinely novel AI considerations, investing in foundational research and strengthening commercial deployment. China joined the non-binding statement. On 2 September the United States also urged G20 members to protect creators while allowing AI training on copyrighted work under fair-use principles, without specifying compensation or implementation rules.
Why it matters: The final text replaces an incomplete policy signal with a documented international position, while the copyright statement exposes a concrete area where national implementation and legal disputes will determine practical impact.
Financial Stability Board Chair Andrew Bailey told G20 finance ministers and central-bank governors that frontier AI could materially change the speed, scale and economics of cyber attacks. The letter links this risk to concentrated technology providers and calls for safer model release, stronger vulnerability management, response and recovery, and resilience among critical third parties.
Why it matters: A global financial regulator is moving the discussion from isolated model incidents to correlated operational risk. Shared models, clouds and service providers can turn one capability or control failure into a system-wide confidence problem.
Independent chairs working with the EU AI Office published the third draft of the General-Purpose AI Code of Practice, covering transparency, copyright and systemic-risk duties for model providers.
Why it matters: The draft translated broad AI Act duties into proposed operational commitments that providers and downstream customers could begin mapping.