Artificial intelligence could lower the barriers keeping biological weapons out of terrorists’ hands, Britain’s MI5 chief warned Tuesday.

Sir Ken McCallum’s warning came with an important qualification: that nightmare scenario has not been realized. His warning concerns what increasingly capable AI could make possible.

That puts a hard question in front of governments and the technology industry: How do you preserve the benefits of powerful AI while keeping it from becoming an assistant to people planning mass murder?

Watch McCallum explain why intelligence services regard increasingly capable AI as an unfamiliar security challenge:

In his annual threat update published by MI5, McCallum said terrorists have already tried using AI to improve surveillance and attack planning. He also flagged the danger of AI-enabled drones.

ADVERTISEMENT

The biological concern involves what more capable systems might enable, rather than a disclosed successful attack. McCallum announced that an AI Threat Assessment Centre had been launched inside MI5, bringing together intelligence, AI safety, law-enforcement and cybersecurity expertise.

Its job includes distinguishing credible dangers from exaggerated claims. That distinction matters when the subject is this serious.

A warning about future capability should trigger preparation. It should not be inflated into a claim that terrorists already possess a weapon the intelligence chief did not say they possess.

The broader security picture helps explain the urgency. This report describes the sharp growth in the hostile-state cases landing on MI5’s desk:

LBC reported that MI5’s work against state threats has quadrupled over the past decade. McCallum said those operations now roughly match the scale of its counterterrorism mission.

He also elevated the agency’s work protecting the country’s resilience, with attention to infrastructure, society and technology. Electricity, water and financial systems were among the everyday services discussed.

LBC’s reporting included a warning from cybersecurity specialist Sai Molige: gaining access to an ordinary company network does not automatically give an attacker control over essential machinery. He warned that access can create an option an adversary may try to exploit later.

That is the practical issue underneath the dramatic headlines. A society depends on systems that have to keep working while someone is actively trying to break them.

Protecting those systems takes more than reassuring language about safeguards. Officials should be able to explain what they are testing, where vulnerabilities remain and who is accountable for fixing them.

ADVERTISEMENT

The warning also prompted a debate over national preparedness. Former defense minister Tobias Ellwood raised the possibility of a modern form of national service in this LBC discussion:

That proposal is a political response, not an announced MI5 policy. Whether it would help is a separate question from whether the threats deserve serious attention.

For Americans watching this unfold, the useful lesson is straightforward: ask specific questions before accepting sweeping promises. Can critical services withstand disruption?

Are dangerous uses of new technology being tested seriously? Can the public distinguish an observed threat from a worst-case forecast?

Powerful technology deserves scrutiny. So do the government powers invoked to control it.

McCallum has put a serious warning on the record. The standard for the response should be concrete protection, measurable results and respect for the freedoms being protected.

 

Join The Conversation. Leave a Comment.