MI5 Launches AI Threat Center as Chief Warns of Bioweapons
MI5 is spending more on state threats than during the Cold War, launching an AI Threat Assessment Center to fight supercharged attacks.
Policy · Source: Reuters
What happened
The UK domestic security service MI5 just officially classified AI as a national security weapon. Director-General Ken McCallum announced the creation of a new AI Threat Assessment Center during his annual speech in London. He warned the public that artificial intelligence is already supercharging cybersecurity threats. The stability of the post Cold War years is gone. The UK has entered a new, highly contested era of national security.
The threat landscape has shifted massively in recent years. McCallum stated that terrorism and state threats are now operating on an equal scale. Hostile foreign states like Russia, Iran, and China are driving this aggressive new era. In fact, MI5 is now dedicating significantly more resources to countering state threats than it did during the height of the Cold War. Over the last decade alone, the agency state threats caseload has quadrupled. Russia remains the number one aggressor, executing hybrid warfare campaigns across Europe.
The most severe warning from the intelligence chief centered on biological warfare. McCallum noted that if governments fail to place sufficient guardrails around AI, the technology could help low sophistication actors create destructive bioweapons. While this specific catastrophic threat has not yet been realized, the danger requires immediate attention. Hostile actors are already utilizing technology to plan drone attacks and sabotage infrastructure. For example, a man was recently charged for allegedly preparing to sabotage a drone factory in Swindon.
Key facts
- Quadrupled — Increase in MI5 state threats caseload over the last decade
- 1,000 — Russian foreign staff expelled from host countries since 2022
- 50 — People arrested in connection with Iran linked plots
Why it matters
For founders building AI products, government scrutiny is about to skyrocket globally. When top tier intelligence agencies launch dedicated AI threat centers, aggressive regulation and strict compliance mandates follow closely behind. Builders working on open source models or dual use technologies will face intense pressure from lawmakers. You will have to prove your safety guardrails actually work against malicious actors. The days of launching powerful models without rigorous security checks are coming to a rapid end.
The second order effect is a massive, sustained influx of capital into AI cybersecurity and defense tech. As state actors like Russia use hybrid warfare and pay local criminals to execute cyberattacks, national defense budgets will swell. Companies that build AI tools to detect, assess, and neutralize these supercharged threats will find eager, deep pocketed buyers. Both the public sector and private enterprise will need new software to defend against AI enabled espionage, sabotage, and coercion.
For builders
Government contracts for AI defense will surge
Intelligence agencies desperately need tools to counter AI driven cyberattacks and hybrid warfare. Founders who pivot to AI security and threat detection will capture massive defense budgets. The public sector is paying top dollar to outpace state adversaries.
Strict model guardrails are no longer optional
MI5 explicitly called out the lack of guardrails as a direct path to bioweapons. Expect aggressive compliance requirements for any model that touches chemistry, biology, or code generation. Startups that ignore safety protocols risk being shut down entirely.
Open source models face severe regulatory risk
Low sophistication actors rely on accessible tools to launch attacks. Governments will likely push to restrict open models to prevent them from falling into the hands of terrorists. Builders relying entirely on open source foundations should prepare contingency plans immediately.
My take
When the head of MI5 says AI is supercharging terrorism, the era of moving fast and breaking things is officially over. Governments are terrified of AI generated bioweapons and will regulate this space with an iron fist. If you are building AI, you are now operating in a national security environment whether you like it or not. Act accordingly.
Original reporting: Reuters. This is my rewrite and opinion.