Essential reading UK law & accountability
Automated decisions. New rules. Human safeguards.
The UK’s revised framework expands when automated decisions can be made. Rights to challenge them still matter.
The briefing
Worth your attention.
Sources reviewed
Artificial intelligenceResearch brief
An agent’s instructions are not its security boundary
AISI’s simulated supply-chain tests put the focus on enforceable permissions and the limits of an evaluation.
Artificial intelligenceNews brief
Justice AI moves into its next phase — with a new focus on risk
The Ministry of Justice reports wider AI use and adds a priority on AI-enabled crime and manipulated content.
Artificial intelligenceNews brief
The FSA puts AI-assisted intelligence to the test
A food-safety pilot is testing whether AI can improve discovery and analysis of regulatory evidence.
Data & privacyNews brief
Smart glasses bring privacy into the everyday
The ICO calls for safeguards that work when AI-enabled recording becomes mobile and less visible.
UK law & policyNews brief
Healthcare AI: oversight beyond the first approval
A UK commission recommends staged authorisation and continuing scrutiny of AI-enabled medical devices.
Artificial intelligenceResearch brief
Better AI evaluation starts with knowing when to stop
AISI’s open-source optstop tool targets unnecessary evaluation runs while tracking statistical uncertainty.
Artificial intelligenceResearch brief
A benchmark score is only as good as the route to it
AISI’s evaluation work shows why task completion and legitimate task completion need separate scrutiny.
Beyond the headlines
The context that counts.
Artificial intelligencePolicy watch
AI tutoring needs supervision — and reassessment as tools change
Newly updated advisory-panel minutes highlight evidence, teacher oversight and clear conditions for using AI tutors.
Data & privacyExplainer
Buying recruitment AI: ask for evidence before the demo
Start with the decision, the candidate and the data flow. A polished shortlist is not a complete procurement case.
UK law & policyPolicy watch
What is next for the ICO’s AI and ADM guidance?
A live guidance roadmap separates work in progress from final regulatory guidance.
Automated decisionsExplainer
An AI agent cannot take your accountability with it
The ICO’s agentic-AI analysis puts organisational responsibility and data access at the centre of governance.
Evidence before opinion
Go beyond the headline.
Go to the source.
Every briefing links to the original publication. We distinguish research from regulation, proposals from law, and our perspective from the source’s findings.
How we choose our sourcesThe reading archive
Older stories. Lasting context.
Earlier research and policy records stay accessible, with their dates and sources intact.