Model releases, frontier-lab strategy, AI policy and the economics of machine intelligence, with a UK lens.
The CMA's technology-neutral proposal could turn ChatGPT-style services into default search providers on Android and Chrome. Google would make every six-monthly eligibility decision under CMA oversight.
An MHRA-appointed commission wants staged access, real-world monitoring and stronger enforcement. Delivering the package would require government adoption, funding and, for some measures, legislation.
EU-facing teams gained 16 months on Annex III conformity work. PRA model-risk expectations and FCA outcomes duties did not move.
The NCSC's 2026 guidance on agentic AI reads like network security. Read closely, it is the nearest thing Britain has to a deployment gate for autonomous AI, and it is being written in controls rather than statute.
The AI Security Institute's own evaluations show frontier models completing half of apprentice-level offensive-cyber tasks and clearing expert-level work for the first time. UK supervisors now measure operational resilience against a threat that is measurably accelerating.
Buried in the Mills Review's seven recommendations is the one with teeth — pull AI systems inside the regulatory boundary and supervise the models, not only the banks that run them.
NCSC-style guidance is telling firms to reserve full agentic autonomy for low-risk prototypes and tighten oversight the moment auth, credentials, or sensitive data enter scope.
The BoE, FCA and Treasury have not written a new AI rulebook; they have made existing resilience duties answer to AI-speed vulnerability discovery, third-party exposure and recovery evidence.
The regulator’s second AI Live Testing cohort points to a harder standard for banks and fintechs: live controls, measurable outcomes and governance that survives real workloads.
The Bank of England, FCA and Treasury are telling regulated firms to treat frontier-model cyber risk as an operational resilience problem now, not as a future AI policy debate.
The Starmer government has unveiled its most ambitious AI investment package to date — £24.25 billion in private commitments, a new South Wales data centre corridor, and a state-backed fund to nurture British AI chips.
The government's £500m venture fund has moved at a pace Whitehall rarely matches, co-investing in Ineffable Intelligence's $1.1bn raise alongside Sequoia, NVIDIA and Google — and signalling that keeping frontier AI companies domiciled in the UK is now an explicit industrial strategy.
OpenAI's GPT-5 arrives with a one-million-token context window, native multimodality, and benchmark scores that leave its predecessors in the dust — but the real story is what it means for AI's role in complex knowledge work across British law firms, banks, and public services.
A trio of competing quantum research teams have independently demonstrated error-corrected logical qubits that outperform physical qubits — the milestone the industry has been chasing for a decade.
As hyperscale AI infrastructure expands at breakneck pace, the world's power grids are buckling under demand — and tech giants are racing to nuclear power as a solution.
With Salesforce reporting 5,000+ enterprise deployments and Microsoft counting 85,000 organisations running custom AI agents, the autonomous workforce is no longer a thought experiment.
With Cursor valued at nearly $10 billion and GitHub Copilot embedded across 50,000 organisations, the AI coding assistant market is racing toward consolidation — and the developer who hasn't adopted one is becoming the exception, not the rule.
Eight months after the EU's General Purpose AI rules came into force, compliance audits are revealing a harsh truth: the gap between what the law demands and what companies can actually deliver is wider than regulators anticipated — and UK firms selling into the EU market are learning that lesson in real time.
