MI5 chief warns AI could 'change the stakes' on bioweapons

The head of MI5, Sir Ken McCallum, has warned that artificial intelligence has the potential to "change the stakes" on existing and even new bioweapons. He delivered the warning in his annual speech on threats to the nation, the yearly address in which the director-general of the United Kingdom's domestic intelligence service sets out the security picture facing the country.

The remark places AI-enabled biological weapons inside the same threat assessment framework MI5 uses for states, terrorism and espionage. That is the core of the story: the UK's intelligence leadership now treats the AI-bioweapons intersection as a standing national security concern, not a speculative one.

The Guardian examined the warning in depth on October 14, 2026, in a report titled "Could AI 'change the stakes' in the development of bioweapons?", and followed with a podcast episode published October 20, 2026, under the title "'These risks are real': the threat of AI bioweapons". The phrase in the podcast headline — "These risks are real" — frames the discussion that follows.

To explain what sits behind the director-general's language, The Guardian's science editor Ian Sample spoke with Dr. Tom Inglesby, director of the Johns Hopkins Center for Health Security and the Bloomberg School of Public Health. Inglesby has spent his career on exactly this territory: the intersection of infectious disease, deliberate biological threats and public health preparedness.

What did the MI5 chief actually say?

Sir Ken McCallum used his annual threats speech to state that AI could "change the stakes" on existing bioweapons — and on new ones. Three elements of that formulation matter.

• The warning comes from the head of MI5, an institution that speaks rarely and in measured terms.
• It names AI as a factor that could alter the risk calculus of biological weapons, not merely accelerate familiar ones.
• It extends to new bioweapons, which signals concern about capability creation, not only the enhancement of known threats.

An annual threat speech from the MI5 director-general functions as an official ranking of dangers. Inclusion in it carries weight across government and informs how resources and attention are allocated.

Why does this matter now?

The podcast episode sets out the question that follows from McCallum's warning: what, specifically, has governments so worried? Dr. Inglesby walks through the concerns in the conversation with Sample.

The stakes are structural. Biological weapons sit at the junction of science, security and public health, and any technology that lowers barriers in the life sciences changes the security equation at the same time. When the head of a major intelligence service says AI could change the stakes, he is signaling that governments now assess the technology as a variable in biosecurity, alongside the traditional concerns of proliferation, state programs and non-state actors.

The October 2026 timing also matters. The warning and the accompanying analysis arrive as governments weigh how to regulate frontier AI systems without suffocating beneficial research, and biosecurity has moved to the foreground of that debate. A threat assessment from MI5 gives policymakers a concrete anchor: the risk is now part of the official picture.

What could reduce the risks?

Inglesby does not stop at the threat description. In the conversation, he sets out what measures could be put in place to reduce the risks — the second half of the podcast's mandate. The episode, running under the blunt headline "These risks are real", pairs diagnosis with response.

The framing tracks the way biosecurity specialists typically approach the problem: identify where AI could plausibly increase danger in the development or use of biological weapons, then design safeguards targeted at those specific points rather than at the technology as a whole. Inglesby's position — director of a dedicated Center for Health Security within a major school of public health — makes him one of the most-cited voices in that effort.

Who is sounding the alarm?

The warning carries two layers of authority. The first is institutional: MI5, the UK's domestic intelligence service, whose director-general delivers the annual threat assessment. The second is scientific: the Johns Hopkins Center for Health Security, which studies pandemics, deliberate biological events and health security policy.

Ian Sample, The Guardian's science editor, conducts the interview. The podcast is part of The Guardian's science podcast series, available at theguardian.com/sciencepod.

The convergence of the two voices — an intelligence chief and an academic health-security director — is itself the signal. When threat assessment and epidemiological expertise point at the same risk, the usual gap between security alarm and scientific caution narrows.

What happens next?

The podcast closes on the practical question its title poses: if these risks are real, what do governments do? McCallum has put AI and bioweapons into the annual threat ledger; the task now falls to regulators, security agencies and the research community to decide which measures get built.

For readers tracking AI policy, the episode is a concise statement of where the biosecurity debate stands after the MI5 warning — and a reminder that the next round of AI safety rules will have to answer a question that until recently belonged to thriller fiction: who can be trusted with tools that might design a pathogen.

Listen to the full episode, "'These risks are real': the threat of AI bioweapons", at theguardian.com/sciencepod.