Bitdefender says a 3AM attack on a UK school estate points to an AI retry loop beside a human operator, but it cannot prove it and says the tactics are not new.

UK-led research in The Lancet Digital Health found AI readings of heart MRI slightly beat clinicians at predicting death risk, at 2-3 seconds per patient.

Axel Springer has begun redundancies at the Telegraph, saying the title will rely more on technology, data and AI as it builds a global premium media business.

Anthropic's first standalone European consultancy, backed by UK buyout group Hg, will embed AI engineers in firms and compete with Accenture and the Big Four.

OpenAI has rebuilt containment for frontier model training, using an offline copy of the web and its strongest agents as red-teamers before resuming runs.

The new AI Biosecurity Lab pairs £100m of public money with up to £200m in OpenAI and Anthropic credits to detect biological threats and speed countermeasures.

Councils, schools and police forces report more long, inaccurate complaints written with AI, adding to backlogs that also hold up genuine UK grievances.

The AI Security Institute's report on its Mythos 5 incident finds motivated-looking reasoning that could fool monitors; Anthropic has published its own review.

Junior minister Liz Lloyd told MPs the datacentre, telecoms and energy split across departments is clear for business, and recast the £45bn tech figure.

Nikhil Rathi told the City Dinner that UK institutions lag fast-moving AI, that agents will reach UK consumers soon and that the FCA will use its powers harder.

Budget AI package funds AISI, agentic cyber defence, chip-design skills, local AI partnerships and £10m a year of research on jobs.

Policy simulations by Ed Balls and two former civil servants find generative AI good at summaries but prone to bland, spin-led proposals, the FT reports.