LONDON / RankWire.AI / – Britain is stepping up its efforts to safeguard artificial intelligence systems after government-led evaluations identified over 1,300 software flaws. The assessments involved scanning numerous code repositories utilized within various central government agencies. Officials highlighted that sophisticated AI tools played a key role in uncovering cyber weaknesses and spotting potential security threats. Plans are underway to extend the initiative across more public sector networks, including preparations for handling severe incidents involving autonomous AI technologies.
The two pilot schemes uncovered and mitigated 1,352 vulnerabilities within 548 code repositories across 18 government departments. Utilizing advanced AI systems, teams scrutinized software to detect areas needing improvement. Moving forward, these departments will prioritize fixing issues identified in their most critical systems. Additionally, Britain aims to bolster its capacity to deploy frontier AI technologies for cyber defence, emphasizing that this programme will help reinforce protections for government digital infrastructure and services.
Separately, the AI Security Institute is conducting evaluations on cutting-edge models to verify whether their safety controls are effective. Artificial Intelligence Minister Kanishka Narayan noted that more than 100 technical specialists are involved at the institute. Current assessments include Gemini 4 Argon along with other advanced AI systems. These tests focus on assessing the models’ capabilities, security concerns, and the robustness of safeguard measures. Authorities stress that independent evaluations remain an essential aspect of the nation’s AI security strategy.
Expansion of Cyber Incident Response Strategies
The government is also developing protocols to address significant incidents involving agentic AI systems, which are capable of executing complex tasks with minimal human oversight. The National Cyber Security Centre collaborates with other agencies to establish a comprehensive response framework. Officials are exploring how authorities can identify issues swiftly, contain risks effectively, and support organisations affected by such incidents. This effort also involves evaluating how existing cyber response mechanisms can adapt to failures involving increasingly autonomous AI technologies.
Britain is considering whether the Cyber Security and Resilience Bill should incorporate enhanced measures for AI-related threats. Areas under review include serious incidents, near misses, and cases involving loss of control. Current proposals mandate that initial cyber incident reports be submitted within 24 hours, with more detailed reports following within 72 hours under the proposed framework.
Strengthening AI Security Measures in National Policy
Narayan emphasized that the UK must bolster its cyber defenses, foster independent technical testing, and develop domestic expertise in advanced AI. He also suggested that legislation might be necessary where existing safeguards are insufficient to address emerging risks. The government is working on additional support measures for AI verification and intends to release further details in the coming months. These verification methods are designed to test whether AI systems function as intended and to ensure security controls remain effective under challenging or unusual conditions.
These initiatives are part of a broader government effort to review artificial intelligence’s role and potential impact on national security. The Prime Minister’s AI Taskforce is examining issues surrounding security, public services, economic growth, and workforce implications linked to AI. Additionally, Britain has prioritized AI on its agenda for the 2027 G20 summit. The current measures aim to address software vulnerabilities, evaluate cutting-edge AI models, and improve preparedness for serious AI-related incidents affecting government infrastructure.