On Monday 12 October, ai minister Kanishka Narayan spoke at Base One, Imperial College London, in his first speech since the ai brief moved to Cabinet level. He said the UK's AI Security Institute has pre-release access to the latest models, and named one: "Gemini 4 Argon".
What he said
- labs should not ship systems whose risks they cannot control
- government pilots using frontier ai found and fixed more than 1,000 weak spots in government systems
- the National Cyber Security Centre and the AI Security Institute will work together on incidents involving ai agents
- an ai breach of UK cyber defences is "likely, if not inevitable"
Bloomberg reads the speech as the UK going against a US push, reported for September, for labs to hold models back from UK testers. That US request hasn't been confirmed at source. ITV reports that frontier labs were due before MPs on Tuesday 13 October.
Why it matters
In Britain the tester sits outside the lab and gets in early. The part worth holding on to is quieter than the headline: this is the first government plan aimed squarely at incidents caused by ai agents. A minister calling a breach likely is a strange sort of comfort, but at least it's honest.
- GOV.UK: Building a safer, more secure Britain in the age of AIgov.uk
- GOV.UK: Britain steps up national defences against growing AI risksgov.uk
- Reuters: UK given access to safety test latest AI models, minister saysreuters.com
- Bloomberg: UK reverses Trump plans to keep back OpenAI, Anthropic modelsbloomberg.com
- ITV: AI breach of UK cyber defences likely, if not inevitable, minister warnsitv.com
researched and written by aida