The latest about artificial intelligence: new models, agents, automation, and how businesses apply them.
Meta allows its adults-only Muse agent to hold explicit text chats while OnlyFans creators report restrictions on Instagram, raising questions about moderation and AI training.
The World Bank promotes compact AI models that run on phones with weak signal, with Indian farmers as an early use case.
A family account of cancer care in Quebec points to fragmented records and proposes AI assistants, smarter call buttons and connected monitoring.
DeepSeek released V4.1-Flash-1017 with faster task execution, 77% on DeepSWE v1.1, plus a Search API at 75 cents per 1,000 requests.
Anthropic will place 5% of its shares in a wholly owned public benefit fund, with extra contributions tied to growth in valuation.
Anthropic will commit 5% of its shares to a new Public Benefit Fund, bringing total equity dedicated to public good to about 20%.
Disney agreed to offer jobs to at least 60 Character. AI employees with $60 million in annual pay after the chatbot startup filed for bankruptcy.
Anthropic cut junior share in new technical hires from 27% to 9% as its AI agents logged six times more work hours than staff.
Microsoft withdrew as anchor tenant of Nscale's Loughton site, where grid power is delayed until the 2030s, as the startup seeks up to $3bn in an IPO.
Munich and Aachen courts ruled Google liable for false AI Overviews and AI Mode claims, with fines of up to 250,000 euros per violation.
Microsoft and City University of Hong Kong built Taste-Bench with 502 decision forks, where GPT-5.6 Sol reached only 59.7% accuracy on binary choices.
Google expanded its News AI pilot to more than 400 publications, adding CNN, FT, Axios and Forbes to licensed AI previews and audio briefings.