OpenAI urges UK to legislate binding AI safety for frontier models
OpenAI's Europe policy chief urged the UK to pass binding, capability-based safety rules for frontier AI, including compulsory third-party testing and strong cybersecurity, with incident monitoring. The proposal aims to align the UK with EU and US frameworks, elevate the AI Safety Institute (AISI) to a central role, and establish flexible, internationally interoperable testing standards to govern rapid AI advances.