MI5 Launches AI Threat Centre, Names Alignment as a Security Concern

MI5 has launched an AI Threat Assessment Centre, Director General Sir Ken McCallum announced on Tuesday in his annual threat update at Thames House in London. The unit, known as AITAC, sits inside MI5 and draws on experts from the AI Security Institute, the National Crime Agency and the National Cyber Security Centre to produce all-source assessments of AI threats to national security.

The centre follows a pattern MI5 has used for new threats: the Joint Terrorism Analysis Centre after the September 11 attacks, and the Joint State Threats Assessment Team about a decade ago. McCallum said AITAC's job is to separate real risks from exaggerated ones, so ministers can make decisions on AI quickly.

He described AI threats that are already active. AI is intensifying the contest between cyber attackers and defenders, he said, and can generate propaganda and radicalizing content tailored to individuals. MI5 has caught terrorists trying to use AI to improve their reconnaissance and attack methods, with AI-enabled drones adding a further problem.

The risk he weighs most heavily has not yet materialized. McCallum said AI could change the stakes on existing or new bioweapons, and Reuters reported that he named terrorists developing bioweapons or even creating new pathogens as his greatest concern.

He then went beyond misuse by people to the risks posed by AI systems themselves, which he identified as the alignment problem.

"Human history has given us no experience of facing an actor which is smarter than we are," McCallum said, adding that the issue deserves the attention it is getting, including inside MI5.

He also cautioned that his concern was measured. "My adrenaline thresholds are set pretty high; I'm not easy to alarm, and neither is my organisation," he said.

McCallum said MI5, GCHQ and MI6 are working closely with the AI Security Institute, and that AI is increasingly speeding up its own development. Separately, he said MI5 warned UK universities two weeks ago that an institute used by Chinese intelligence has been funding UK research on topics including AI and cybersecurity, and he told tech startups to expect foreign spying.

MI5 did not say when in 2026 AITAC began work, how many people it employs, who leads it, or whether any of its assessments will be published.

The speech comes as the AI Security Institute features prominently in the AI safety debate. Microsoft President Brad Smith called it the world's strongest AI testing body this week, as AI Briefing News reported. AITAC's reliance on the institute links the UK's AI testing work directly to its intelligence services.

McCallum delivered the update as MI5's workload rises across the board. He said MI5 now spends more resources countering state threats than it did during the Cold War, and called Russia "aggressor number one." Britain in April raised its national terrorism threat level to "severe", meaning an attack within the next six months is highly likely.