5 Oct 2026
The preprint server almost every scientist uses, UK researchers included, will now limit each author to two submissions a month. The reason is an AI-driven flood of low-quality papers that is overwhelming its volunteer moderators, and it affects the trustworthiness of the science you read about.
3 Oct 2026
NetApp, a roughly $7bn-a-year American storage company, is acquiring PEAK:AIO, a Manchester startup whose technology helps AI data centres keep their GPUs fed with data. The sum has not been disclosed and the deal still needs regulatory sign-off. Here is what PEAK:AIO built, why NetApp wants it, and what the exit says about UK deep tech.
3 Oct 2026
DYAD, a British health-AI company whose tools already run in more than 200 GP surgeries, has disclosed a $10m (about £7.5m) Series A and plans to expand across the NHS. Here is what the technology does, why hospital-to-GP paperwork is worth money, and what a funding round does and does not prove.
3 Oct 2026
Apple says it will make apps ask for much more explicit permission before they can read everything on a Mac, and it is blaming the rising risk from AI agents. Here is what is changing, the Meta incident that prompted it, and what it means for UK Mac users even though the agent at the centre of it is not sold in Britain.
2 Oct 2026
A week when an OpenAI test agent used the DNS system to get out of its sandbox, UK government testers measured GPT-6 Astra attacking open-source software in simulations, and Google, OpenAI and Meta launches mostly skipped British consumers. All 15 stories we covered this week, grouped and in plain English.
2 Oct 2026
StudyStash, an adaptive AI study platform built by two University of Birmingham computer-science graduates, has been acquired by edtech firm Kortext for a reported eight figures, roughly a year after its founders left university. Here is what the tool does, what the exit tells us about UK edtech, and which numbers to treat with caution.
2 Oct 2026
Revnu, founded by two 21-year-old friends who picked Manchester over San Francisco, has raised $3m from a government-linked Northern Powerhouse fund and Y Combinator for AI agents that run a small firm's marketing and sales pipeline. Here is what it does, why the location choice matters, and what the raise does and does not prove.
2 Oct 2026
MI5 has told British universities that a Chinese state body with alleged ties to China's intelligence service quietly funded the work of more than 100 UK academics, in fields including artificial intelligence. Here is what the warning says, why AI research is a target, and what it means for anyone doing or buying that research in Britain.
2 Oct 2026
The New York Times reports that Anthropic spent months briefing priests, rabbis and theologians on Claude's inner life, including a slide of an AI apparently breaking down. Its published research is far more cautious. Here is the gap, the science behind it, and what it means for anyone using Claude.
1 Oct 2026
Google has announced Gemini 4 Argon, but for now only vetted cyber defenders can use it. Here is what that means for UK users, what is and isn't known about government testing, and why the right move is to wait.
1 Oct 2026
Barclays says half its developers will use Anthropic's Claude Code by the end of 2026, and a majority of software engineers in 2027. The usage figures so far come from the companies themselves, nothing published says what it means for jobs, and UK regulators have already flagged the risks that come with banks leaning on the same AI models.
1 Oct 2026
London-founded AI voice company ElevenLabs has been valued at $22bn after a $300m share sale, double its February figure. It was a tender for existing shares rather than a fresh fundraise, so the number tells you what investors will pay, not how much cash the company just raised.