LONDON / RankWire.AI / – Britain is bolstering its defenses against the risks posed by advanced AI after trials uncovered over 1,300 cyber vulnerabilities. These weaknesses were found across hundreds of government software repositories. The British government stated that its initiative will leverage frontier AI technologies to improve cyber security and pinpoint vulnerabilities within critical digital infrastructure. Additionally, officials are developing plans for serious incidents involving autonomous AI systems. This effort is part of a broader strategy to reinforce national resilience as more sophisticated AI models become accessible to the public.
During two government-led pilots, 1,352 vulnerabilities were identified and remedied across 548 code repositories spanning 18 departments. The experiments aimed to assess how advanced AI could assist in locating software flaws and recognizing potential cyber threats. The authorities now intend to expand the frontier AI hardening initiative across government agencies. Focus will be placed on systems with the highest risk profiles. The program also seeks to develop broader government capabilities in deploying AI for cyber defence while maintaining oversight of these technologies.
Britain’s AI Security Institute is currently evaluating cutting-edge models and testing their safeguards’ effectiveness. Artificial Intelligence Minister Kanishka Narayan highlighted that the institute employs more than 100 technical specialists dedicated to model assessment and security. Their ongoing work includes evaluations of Gemini 4 Argon and other advanced AI systems. The government emphasizes the importance of independent testing, noting that public authorities cannot solely rely on AI developers to evaluate the risks associated with increasingly capable models.
Expansion of AI incident response planning
National Cyber Security Centre collaborates with government bodies and the AI Security Institute to establish protocols for major incidents involving agentic AI. These autonomous systems are capable of executing complex tasks with minimal human oversight. The planned framework will serve as a guide for authorities to detect incidents, contain risks, and support those affected. Officials are also reviewing how existing cyber response mechanisms can handle failures involving autonomous AI and other advanced technologies in vital services.
Additionally, the government is contemplating amendments to the Cyber Security and Resilience Bill progressing through Parliament. Proposed measures include enhanced management of AI-related cyber risks, serious incidents, near misses, and events where control is lost. Current proposals mandate that regulated operators and digital service providers report significant cyber incidents. Operators would be required to submit initial notifications within 24 hours, followed by comprehensive reports within 72 hours. The security framework encompasses critical sectors such as hospitals, energy infrastructure, and drinking-water systems.
Britain advances AI security protocols
Narayan stated that Britain must strengthen its cyber defenses through independent testing and domestic technical expertise. He also mentioned that legislation might be necessary if existing safeguards do not sufficiently mitigate new risks. The government is actively developing support systems for AI verification, with plans to share further details soon. These verification tools are intended to confirm that advanced systems behave as expected and that safety measures remain effective under demanding conditions.
These initiatives form part of a larger government effort to examine the national security and public service implications of artificial intelligence. The Prime Minister’s AI Taskforce is investigating issues related to economic growth, public services, the workforce, and security. Britain has also designated AI as a key policy area for its 2027 G20 presidency. The latest cybersecurity efforts concentrate on testing advanced models, fixing software vulnerabilities, and preparing government systems for serious AI-related incidents.