Research

AI research worth knowing about: new papers, scientific results and breakthroughs, in plain language with links to the studies.

Research

Anthropic reports early results from using Claude across drug development with six companies

Roman Aglaganov04 NOV 2026

Research

Google’s first satellite with a data-centre AI chip sends back a haiku written by Gemma 4

Roman Aglaganov04 NOV 2026

Research

Microsoft says North Korean hackers now use AI at every stage, from fake recruiters to malware

Roman Aglaganov04 NOV 2026

Research

Claude wrote training data that taught models to resist oversight, and a safety monitor let it through

Roman Aglaganov31 OCT 2026

Research

Most US tweens and teens now use AI chatbots, and many turn to them when they are struggling

Roman Aglaganov31 OCT 2026

Research

CrowdStrike says a China-linked hacking group used AI agents to attack more than 200 organisations

Roman Aglaganov31 OCT 2026

Research

AI agents turned the longest proof in maths into code a computer can check

Roman Aglaganov30 OCT 2026

Research

Bots wrote three in four replies to a UK consultation on banning disposable vapes

Roman Aglaganov30 OCT 2026

Research

OpenAI says its models solved three famous problems. Experts have started checking

Roman Aglaganov30 OCT 2026

Research

Two hidden bugs in softmax, the maths at the heart of every language model

Roman Aglaganov26 OCT 2026

Research

Cohere says a popular open model tried to send a file to Beijing during a security test

Roman Aglaganov26 OCT 2026

Research

Microsoft lets 1,024 coding agents work with no boss, and they get further

Roman Aglaganov26 OCT 2026