G20 backs lighter AI rules as Sanders seeks superintelligence ban

AI governance moved in sharply different directions across major jurisdictions between September 1 and September 3. The G20 endorsed the US-proposed Carolina Principles, a non-binding framework that favors sector-specific AI oversight through existing authorities and discourages new AI-specific regulatory bodies. All twenty member nations backed the approach, though Demis Hassabis publicly called for a FINRA-style body to test powerful systems before release.

Senator Bernie Sanders and Representative Greg Casar introduced the Ban Artificial Superintelligence Act one day later. The bill would permanently ban systems that match or exceed human cognitive performance across broad domains, or that can subvert shutdown commands, and would pause advanced AI development until a new cabinet-level agency sets safety and model-review standards. Penalties include up to twenty years in prison for individuals and forced dissolution for companies. The measure listed no cosponsors or endorsing organizations and is unlikely to advance this session.

California sent thirty AI-related bills to Governor Newsom, shifting from a failed broad frontier-model proposal in 2024 to targeted rules covering workplace surveillance, auditor registries, children’s chatbot safety, student privacy, and deepfakes. In Europe, regulators in France, Germany, and Spain began the EU AI Act’s first on-site audits for high-risk systems deployed after August 2, with violations carrying fines of up to fifteen million euros or three percent of worldwide turnover.