UK Committee Holds 1 AI Security Hearing as Ministers Warn of Hacking and Bioweapon Risks
Updated
Updated · The Guardian · Oct 17
UK Committee Holds 1 AI Security Hearing as Ministers Warn of Hacking and Bioweapon Risks
2 articles · Updated · The Guardian · Oct 17
Summary
Liam Byrne’s parliamentary committee held an evidence session this week with civil servants and executives from OpenAI, Anthropic, DeepMind and Meta as Westminster sharpened its focus on AI security.
Monday’s warning from AI minister Kanishka Narayan that rogue agents hacking the British state was “likely, if not inevitable,” followed by MI5 chief Ken McCallum’s alert on AI-enabled bioweapons, helped drive the shift.
Byrne said witnesses split over whether government testing of new models should be mandatory, while the UK is also trying to preserve company cooperation with its AI Security Institute despite reported pressure from Donald Trump’s administration.
Experts told lawmakers open-source models could already do serious damage because safety guardrails can be removed, though some academics argued immediate harms such as hacking and addiction deserve more attention than existential-risk scenarios.
The debate now reaches beyond regulation to UK readiness: defence officials are split over whether safety fears are slowing capability-building, even as the MoD points to a £400 million AI investment over four years.