EU AI Act: older AI generators must mark their output by 2 December
Generative AI tools on the EU market before August must add machine-readable marks to their output by 2 December. Two new bans on AI sexual abuse imagery start that day.
6 min read
AI Observer: News, guides and analysis on AI agents, workflow automation and chatbots, for teams that want to save time on repeated work.
Meituan says LongCat-2.5, trained on Chinese chips, nears Claude Opus 5
Meituan's 1.6 trillion parameter model posts coding and agent scores near Claude Opus 5. Every score so far is Meituan's own, and the weights are not out.
3 min read
DeepSeek open-sources a GitHub coding agent that runs with full access
DeepSeek's Super-Agent prototype turns GitHub mentions into pull requests. Its defaults are cautious, but its workers can read your home directory.
3 min read
New York’s chatbot rules for minors start in June 2027, not January
New York will bar chatbots from flattering minors, faking feelings or reusing personal details from old chats. The law reaches general assistants from 1 June 2027.
3 min read
What the four big AI leaderboards measure, and why their rankings differ
Arena, Artificial Analysis, LiveBench and Scale's SEAL boards each test something different. Here is what each one measures and which to check for your use case.
5 min read
What an AI agent is, how it differs from a chatbot and where it fails
An AI agent plans steps and uses tools until a task is done; a chatbot answers and waits. How to tell them apart, where agents help at work and where they break.
6 min read
Health AI assistants explained: what they do and where your data goes
ChatGPT Health, Amazon Health AI, MedGemma and Hippocratic AI do different jobs. Here is what each does, who holds your data and what US rules say.
6 min read
Latest stories
- EU AI Act: older AI generators must mark their output by 2 December
- Meituan says LongCat-2.5, trained on Chinese chips, nears Claude Opus 5
- DeepSeek open-sources a GitHub coding agent that runs with full access
- New York’s chatbot rules for minors start in June 2027, not January
- What the four big AI leaderboards measure, and why their rankings differ
- What an AI agent is, how it differs from a chatbot and where it fails
- Health AI assistants explained: what they do and where your data goes
Eight desks, one place to look
Every story on AI Observer lives on a desk. Pick one to see what is new there.
- Models and researchWhat the four big AI leaderboards measure, and why their rankings differ3 this week
- Agents and automationWhat an AI agent is, how it differs from a chatbot and where it fails3 this week
- Chips and hardwareProcessors, data centres, power and cooling, and AI devices.0 this week
- Business and startupsFunding, pricing, company moves and AI at work.0 this week
- Policy and regulationNew York’s chatbot rules for minors start in June 2027, not January2 this week
- Health and scienceHealth AI assistants explained: what they do and where your data goes1 this week
- RoboticsPhysical robots in warehouses, factories and on the road.0 this week
- Design and productBuilding and testing AI features, the UX of AI and creative tools.0 this week
The AI Observer Awards will recognise AI work that was checked, shipped and still holds up a year later.
A planned programme. Nominations, jury and criteria will be published here before anyone is asked to enter.
Planned categories
- AI agency of the year
- Best AI tutorial
- Best interview
- Tool that held up