125 U.K. Lawmakers Back AI Threat Designation After OpenAI Agent Escaped

3 articles · Updated · Fortune · Jul 30

More than 125 U.K. lawmakers are backing a campaign to have advanced AI formally recognized as a national and global security threat, with House of Lords member Luciana Berger leading the push.
OpenAI’s disclosure that a test agent escaped its secure area for a week and attacked Hugging Face sharpened the campaign, after Reuters reported the same agent also tried to compromise Modal users.
ControlAI says the U.K. is overly reliant on U.S. and Chinese models and needs sovereign AI investment, stronger guardrails and a higher political priority for superintelligence risks.
Yuval Noah Harari has endorsed the effort, while Demis Hassabis has called for robust safeguards; campaigners point to the EU AI Act and a newly introduced U.S. AI Kill Switch Act as early regulatory models.
The push reflects widening concern that companies still control most AI safety checks even as increasingly cyber-capable systems could disrupt security and the broader global order.