GOV.UK. DSIT published research on cyber security vulnerabilities in operational technologies, relevant to AI governance where AI-enabled monitoring and control systems sit inside critical and high-risk industrial environments. GOV.UK. The Department for Transport published a detailed evaluation of its AI Consultation Analysis Tool (CAT), setting out accuracy testing against human benchmarks, a human-in-the-loop pilot design, and bias checking using protected-characteristic proxies, with an explicit aim of building trust in government AI use through transparent evidence.

The Law Society. The Law Society gave evidence to the Joint Committee on Human Rights on AI and human rights, pointing to rule-of-law and accountability issues as policy work on AI continues in the UK. European Data Protection Supervisor. The EDPS published Newsletter 117, highlighting EU-level work on preparing public administration for AI risk and AI Act implementation, including an inventory-style mapping exercise of AI systems within EU institutions and broader horizon-scanning on AI trends.

GOV.UK. The Medicines and Healthcare products Regulatory Agency has launched a high profile Regulation of AI in Healthcare call for evidence, presented as a pivotal moment for the UK framework and inviting views on safety checks, liability allocation and post deployment monitoring for AI medical devices. GOV.UK (+AISI). The UK AI Security Institute has released its first Frontier AI Trends Report with an accompanying government factsheet, publishing aggregated test results on more than thirty frontier systems and showing that capabilities in areas such as code generation and biology are advancing rapidly while serious vulnerabilities and misuse risks remain.

Ministry of Justice (MoJ) published an Engineering AI Governance Framework for its engineering teams, setting practical rules for using tools like GitHub Copilot and for building bespoke AI systems across the development lifecycle. GOV.UK. European Parliament adopted a resolution urging the Commission to propose new rules on algorithmic management at work, including transparency, worker consultation and restrictions on intrusive monitoring.