Microsoft Urges UK to Drive Global AI Regulation

Microsoft President Brad Smith is urging governments to establish binding rules for artificial intelligence, naming the United Kingdom as the ideal country to guide international oversight. Speaking on a BBC podcast, Smith warned that AI represents both a tool and a weapon that risks disrupting global cybersecurity, noting that common household items currently face stricter regulatory oversight than frontier technology firms. He pointed to Britain’s AI Security Institute, launched in 2023, as a key player capable of evaluating models before deployment, contrasting its neutral standing against Brussels in the eyes of Washington officials.

The push for British oversight persists despite political friction in the United States. President Donald Trump recently criticized international safety bodies and asked American firms like OpenAI and Anthropic to stop sharing unreleased models with British evaluators, though the institute confirmed its ongoing access to new releases. While White House talks recently led executives from Google, Meta, SpaceX, AMD, Nvidia, Anthropic, and OpenAI to sign a morally binding agreement regarding independent security audits, Smith argued that voluntary measures fall short because bad actors simply ignore non-enforceable pledges.

To protect against rogue autonomous agents—a concern also voiced by leaders like Dario Amodei of Anthropic and Sam Altman of OpenAI—Smith outlined specific policy requirements. These include legally mandated breach notifications, external safety sign-offs prior to public release, and emergency kill switches to pause runaway software. For developers and AI practitioners, these proposals signal a shift away from corporate self-policing toward compulsory external auditing. If implemented globally, engineering teams will need to integrate fail-safes directly into their system architectures and accommodate pre-deployment evaluation protocols from third-party safety bodies.