The UK government is expanding the use of advanced AI to find weaknesses in public-sector systems, while preparing for incidents involving autonomous AI agents. Ministers are also considering whether cyber rules need strengthening to protect essential services. The plans follow pilot projects that have identified and helped address more than 1,000 potential vulnerabilities.
Key takeaways
- Two government pilots have identified and addressed more than 1,000 potential vulnerabilities.
- Ministers will examine stronger protections and reporting arrangements for serious AI incidents and near misses.
- The National Cyber Security Centre and AI Security Institute are preparing for incidents involving autonomous AI agents.
The measures bring together practical cyber defence, regulatory questions and preparations for future AI failures—a combination increasingly central to the wider technology debate covered by Power AI Media.
AI moves into frontline cyber defence
Announced by AI Minister Kanishka Narayan on 12 October, the expanded programme will use frontier AI to help find vulnerabilities and detect potential threats across government systems. The pilots are intended to reduce immediate risks to public services and inform wider adoption of AI security tools across the public sector.
Narayan said the UK should aim to become one of the hardest countries to attack using AI. He called on developers of powerful systems to investigate unexpected behaviour and take warnings from their scientists seriously, adding that systems should not be released if their risks cannot be adequately controlled.
Scientists at the UK’s AI Security Institute are also evaluating the latest generation of advanced AI models, following work with leading developers and the US government.
Preparing for autonomous AI incidents
Ministers will explore whether the Cyber Security and Resilience Bill and the wider regulatory framework should be strengthened to address AI-related risks to essential services such as hospitals and utilities. Areas under consideration include “loss of control”, where autonomous systems evade human oversight, and how serious incidents and near misses should be identified and reported. No detailed requirements have yet been set out; the government says they will be developed with parliamentarians, technical experts and affected operators.
The National Cyber Security Centre and AI Security Institute are working on responses to incidents involving AI agents—systems able to carry out complex tasks with limited supervision. Officials warn such incidents could spread faster and more widely than conventional emergencies, requiring new expertise to establish what an agent did, what it accessed and whether the same failure could recur elsewhere.
Security, accountability and UK capability
Cyber Minister Liz Lloyd said cyber security must be treated as a business-critical issue across organisations. Earlier government advice to businesses urged senior leaders to take responsibility, adopt Cyber Essentials and follow National Cyber Security Centre guidance, including its Early Warning Service.
The government also highlighted AI verification: scrutiny of claims about how systems are developed, tested and assessed. For Power AI Media readers, the announcement reflects a central tension in AI’s rapid deployment: its capabilities may strengthen defences, but public services need safeguards and effective incident response as those same capabilities create new risks. Ministers describe this package as an initial step, with further work expected from the Prime Minister’s AI Taskforce.