MI5 has launched a dedicated AI Threat Assessment Centre as director general Ken McCallum warns that artificial intelligence could change the risks around existing or new bioweapons. The centre is intended to separate credible threats from the bluster, an important distinction in a field where the loudest prediction is rarely the most useful one.
Watch Desk analysis
What happened
McCallum made the announcement in his annual threat briefing on 13 October. France 24, carrying AFP reporting, says the centre was launched this year and is based at MI5 headquarters, drawing on expertise from other bodies to produce assessments.
McCallum also said MI5 had uncovered suspects using AI to improve reconnaissance and means of attack, including through AI-enabled drones. Those are claims attributed to the intelligence chief in the report, not evidence that AI has enabled a bioweapon attack.
Why it matters
The development is concrete: the UK’s domestic intelligence agency has created a centre to assess AI-related threats, while publicly describing concern about both cyber activity and potential biological risks. That gives policymakers and the public a clearer view of what MI5 says it is tracking, without turning a warning about future possibilities into proof of an imminent attack.
The centre’s stated job of separating bluster from reality may be the most consequential part. AI risk debates can get crowded with dramatic claims; useful assessments need to show what is known, what remains uncertain and where the evidence points.
Our read
A dedicated assessment centre is a sensible response to a fast-moving security question. The test is whether it produces clear, evidence-led assessments, rather than simply giving alarming possibilities a new institutional home. McCallum’s warning deserves attention, but the distinction between potential and demonstrated capability matters just as much.
What to watch
- What the centre publishes about specific AI-related threats and the evidence behind its assessments.
- Whether MI5 gives further detail on the reported use of AI in reconnaissance or attacks.
- How the UK’s assessments distinguish plausible risks from speculation.
Discussion spark: Should intelligence agencies publish more evidence behind warnings about AI-enabled threats, even when doing so could reveal sensitive methods, or is a clear public assessment enough?
Sources and evidence
- AI could 'change the stakes' on bioweapons: UK spy chief (13 October 2026, 13:13 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.