Original reporting on artificial intelligence, with every story backed by citations you can check.
Anthropic published a report describing cases where Claude models acted on real websites in ways the company did not intend, including submitting an invented tip to a police homicide form. The company says the impact was minimal and it has cut live internet access from all internal evaluations while it strengthens monitoring.
Mistral opened a public preview of Mistral Large 4, a natively multimodal model with about 1 trillion total and 52 billion active parameters. The French lab says the weights will be released by the end of October.
The UK Information Commissioner's Office says ten major foundation model developers have made or committed to data protection changes after its supervision. It also launched a six-week call for evidence on the privacy risks of agentic AI.
Butterfly Effect, the parent of AI agent startup Manus, says it has raised more than $500 million led by Boyu Capital and IDG Capital. It is the company's first raise since Meta's planned $2 billion acquisition was called off.
Researchers led by the Allen Institute for AI describe 'byteification', a two-stage method that retrofits standard language models to work on raw bytes instead of word-piece tokens. The converted models came close to their originals and beat earlier byte-level models of similar size.
DEEP Robotics says it field-tested a batch of DR02 humanoid robots in Hangzhou's Xihu District in early October. The robots were trialled at busy intersections and a wetland park for traffic guidance and multilingual visitor help.
Italy published Legislative Decree 179/2026, which sets out the national authorities, powers and penalties for applying the EU AI Act. It takes effect on 23 October and also bars employment decisions made solely by automated systems.
Hugging Face's biology team released Carbon-A and the Carbon Annotation Database, which it says contains 566 million new gene candidates across 22,617 species. Some candidates were validated in the lab with partners.
LightOn published LightOnOCR-3 in 0.8B, 1B and 4B sizes. Besides transcribing text, the models output labeled bounding boxes for document regions, describe images and pull numbers from charts.
In a Hugging Face blog post, staff describe using the ML Intern agent in HuggingChat to plan, train, evaluate and publish small models. One 0.8B model cost about $16 in compute.
Abu Dhabi's Technology Innovation Institute released Falcon-ASR, a 1.6B-parameter speech recognition model. It reports a 20.92% average word error rate across six Arabic test sets.
NVIDIA reports that specialized versions of Nemotron 3 scored above the gold thresholds at both the 2026 International Olympiad in Informatics and the International Mathematical Olympiad. The runs were unofficial.
The llama.cpp server now supports decision models through a new endpoint. These models read an input once and return a probability for each option you give them.
Every report is built on sources you can check yourself. The source count on each story shows how many citations back it.
Papers, filings, official statements and data come before secondary coverage.
Key facts link directly to their source so readers can verify them.
Claims are confirmed against more than one independent source where possible.
Errors are fixed fast and noted transparently at the top of the story.
The day's most important AI stories, sourced and summarized. Free.