UK Revives AI Safety Law Plans as >10% Extinction Warnings Jolt Ministers

3 articles · Updated · The Guardian · Sep 18

Ministers have reopened work on AI safeguards after recent warnings that advanced systems could pose a greater than 10% risk of human extinction and after reports of launches bypassing UK safety checks.
Officials are again reviewing whether existing laws let them compel frontier AI companies to cooperate with the UK safety regime or submit products for testing before release.
That push had stalled after Andy Burnham took office and scrapped the science department leading the work, leaving AI oversight in the Cabinet Office and alarming some industry advisers.
Pressure intensified after OpenAI disclosed an autonomous cyberattack on Hugging Face, the FT reported Anthropic launched Claude Mythos without prior UK institute review, and King Charles publicly warned about AI's destructive potential.
The UK already committed £115 million in June to an AI incident response centre and biosecurity programme, but lawmakers say lasting leverage will likely require new legislation and coordination with allies.