The latest about artificial intelligence: new models, agents, automation, and how businesses apply them.
North American startups raised $92 billion in Q3, down 35% from Q2, while AI companies captured $61 billion ahead of OpenAI and Anthropic IPOs.
OpenAI released 372 mathematical results from one internal model on GitHub, each from a single prompt and Lean checks to ease review.
Atlassian unveiled Agentic Multiplayer Protocol at Team '26 Europe, plus Rovo Work and MCP server with 200 tools handling 15 million calls daily.
OpenAI published 722 manuscripts in 372 families from about 4,000 problems, averaging three hours of ChatGPT Pro compute per result.
OpenAI released 722 manuscripts in 372 families from 4,000 attempted problems, with Lean proofs and reasoning summaries in an open repository.
Dell added a semantic layer, knowledge graph and Knowledge Agents to its AI Data Platform, tying processing closer to storage for agentic AI.
MasterClass Executive runs about 10 agents per learner interaction, with 30,000 applications for 500 spots in the first cohort.
Two-year-old Mirror Particle rejects LLM role-play and builds a behavior model from scratch for brand strategy.
MIT launched the MIT for America initiative to scale STEM and AI learning from kindergarten to community college, building on calculus tutoring and RAISE.
Gamma introduced Gamma 5 with a rebuilt editor, agentic AI and thousands of templates to eliminate repetitive layouts in AI presentations.
OpenAI tracked 1.5 million ChatGPT messages and 6,200 workers, as cross-field AI use rose from 13.1% to 25.9% and expert-led writing gains ground.
DeepSeek is close to raising 80bn yuan, or about $12bn, with Tencent and CATL as largest backers and the total possibly near 100bn yuan.