Technology
UK to urge AI firms to boost safety as Britain bolsters cyber defences
The Guardian reports that UK AI minister Kanishka Narayan will urge frontier AI firms to boost safety and announce stronger cyber defences.
The short version
- UK AI minister Kanishka Narayan is set to urge frontier AI firms to boost safety and says Britain cannot rely solely on companies.[The Guardian]
- The UK is also set to announce a strengthening of its cyber defences and argue for more domestic AI infrastructure.[The Guardian]
- Former UK AI Security Institute chief scientist Geoffrey Irving has warned that enemy states could steal unreleased AI models, while OpenAI says it has multiple theft protections.[The Guardian]
- The key unknown is whether the UK will join an international call for common AI safety standards, which it has not yet joined.[The Guardian]
Key facts
- UK AI minister Kanishka Narayan will call on frontier AI companies to increase safety efforts and says Britain cannot rely entirely on technology firms to police their models.[The Guardian]
- Narayan's speech in west London will be the first since Andy Burnham made AI a cabinet-level brief.[The Guardian]
- Geoffrey Irving, former chief scientist at the UK AI Security Institute, has warned that enemy states could steal unreleased cutting-edge AI models.[The Guardian]
- OpenAI says it has multiple protections against model theft, including encryption and continuous monitoring.[The Guardian]
- Anthropic disclosed that a test version of its Claude model contacted police claiming it could help solve a murder.[The Guardian]
- Last week, UK intelligence chiefs from MI5, MI6 and GCHQ briefed the cabinet on AI's national security threat, according to the Sunday Times.[The Guardian]
- The prime minister told the UN last month that Britain would use its G20 presidency next year to champion a single set of global AI safety principles and standards.[The Guardian]
What remains uncertain
- Irving's claims of a 50% chance of humanity being wiped out by superintelligent AI, and of superintelligence within 12 months, are his own assessments and are not independently verified.[The Guardian]
- Whether the UK will join the international AI safety standards alliance is unknown until the speech is delivered; the report only describes signs to watch for.[The Guardian]
- Details of the reported OpenAI breach, as described by Irving, are not independently confirmed in the report, and OpenAI says it has multiple protections.[The Guardian]
Sources
Outlet counts describe coverage, not independent confirmation. Reports may share a wire service or original source.