MI5 has elevated its work to protect the UK against hostile activity to a third core mission, placing it alongside counter-terrorism and countering state threats. Director General Sir Ken McCallum announced the change in his annual threat update at Thames House on 13 October, alongside the launch of a new centre to assess national-security risks linked to artificial intelligence.
The Protect mission formalises and raises the profile of work MI5 says it has carried out for more than a century. The service describes it as its contribution to national resilience: helping make the country a harder target and supporting its ability to keep functioning when incidents occur. The change reflects, in McCallum’s words, a more contested security environment, rather than the creation of a wholly new protective role.
A broader resilience effort
Protect is intended to reach beyond MI5’s direct investigations. The update describes a “team of teams” approach involving government and wider society, with resilience spanning essential systems, communities and technology. Those areas overlap: disruption to infrastructure can affect daily services, while hostile activity can also seek to exploit social divisions or digital vulnerabilities.
MI5’s National Protective Security Authority works with the National Cyber Security Centre on advice for organisations responsible for critical infrastructure. The stated areas include hostile reconnaissance, sabotage, insider risk, drones and cyber threats. The announcement presents this work as part of a wider national effort; it does not set out a new public performance target or claim that the risks can be eliminated.
New centre to assess AI risks
The AI Threat Assessment Centre, or AITAC, has been launched in 2026 and is housed within MI5. It brings together expertise from several organisations, including the AI Security Institute, the National Crime Agency and the National Cyber Security Centre, to produce assessments drawing on multiple sources. Those assessments are intended to help government departments understand emerging threats and make decisions as AI develops.
The centre’s remit is assessment, not a public announcement of a new enforcement power. MI5’s account says the work is meant to separate exaggerated claims from practical risks. Its short public description says it will provide assessments for government departments, but gives no details of staffing, budget, timetable for publications or the process by which particular departments will use its work.
The update describes several ways malicious actors may exploit AI. These include strengthening cyber activity, generating tailored propaganda and improving reconnaissance or attack methods. It also refers to AI-enabled drones. Such examples describe areas of concern, not evidence that every capability is in routine use or that a particular attack is imminent.
McCallum singled out the possibility that AI could change the stakes around existing or new biological weapons as a serious potential misuse case. He said that risk had not yet been realised. The distinction matters: the centre is being set up to assess threats and inform government, not because the announcement establishes that an AI-enabled biological attack has taken place.
Established missions remain
MI5’s two other core operational missions remain countering terrorism and state threats. The new structure adds protective security and resilience as an equal priority, while the assessment centre focuses on the security implications of AI across different kinds of threats. The announcement does not describe AITAC as replacing existing intelligence or operational teams.
For the public, the change signals a greater emphasis on preparedness as well as disruption of adversaries. That involves attention to infrastructure, organisations and social resilience, alongside analysis of technologies that may be used by hostile actors. The practical effect will depend on how advice and assessments are translated into decisions across government and the sectors they cover.
Both developments are framed as responses to changing risks, not guarantees of safety. MI5 says resilience can limit the impact of hostile actions and reduce their appeal, while the new centre is intended to provide a clearer assessment of AI-related dangers. The announcement leaves the details of future implementation to follow.
