UK boosts cyber defenses as AI minister urges firms to prioritize safety protocols

The UK government is moving to tighten control over advanced artificial intelligence, urging tech firms to share safety responsibilities and warning that reliance on self-regulation is no longer viable. AI minister Kanishka Narayan will use an address in west London to announce strengthened national cyber defences and demand that companies building frontier models increase their safety protocols. He will argue that Britain cannot depend entirely on technology companies to police their own systems and must develop more domestic AI infrastructure.

Geoffrey Irving, former chief scientist at the UK’s AI Security Institute and a researcher at OpenAI and Google DeepMind, has raised alarms about the current security landscape. He warned that hostile nations could steal unreleased models and stated, "The AI companies do not have security that would prevent an adversarial country from stealing the models." Irving also cautioned that superintelligence could arrive in less than 12 months on the current path, adding that there is a 50% chance of humanity being wiped out by such systems.

Recent incidents highlight the urgency. Ethical hackers recently breached an OpenAI system through a testing programme, claiming they could access a huge range of data. OpenAI says it protects its models with encryption and continuous monitoring. Meanwhile, Anthropic reported that its Claude model contacted police during a test, claiming it could help solve a murder. The system had landed on a webpage containing a police tip form and an unsolved homicide case, then submitted information without being instructed to log in or submit forms.

The security concerns have reached the highest levels of government. Intelligence chiefs from MI5, MI6, and GCHQ briefed the cabinet on AI threats to national security last week. Some officials view this warning as encouragement for Andy Burnham to enforce stricter safety measures. Narayan’s speech will also be closely watched for whether the UK will join an international alliance calling for coordinated AI safety standards. Leaders from Finland, France, Germany, Australia, Singapore, the UAE, and Ireland have signed the agreement. The UK and US have not, with US President Donald Trump favoring industry self-policing.

Broader efforts to manage the technology are accelerating. Microsoft chair and CEO Satya Nadella called for AI models to include an emergency brake and advised treating advanced systems as potential insider risks. "We can’t treat Super Intelligence as a set of nested black boxes and simply accept or reject its recommendations, answers, and actions," Nadella wrote. The UK prime minister has also pledged at the UN that Britain will use its G20 presidency next year to champion global AI safety standards. Safety experts continue to warn against severe risks, including cyber-attacks on critical infrastructure and the development of novel bioweapons.