Curated AI news from the last 24 hours — models, agents, enterprise, and developer tools.
Mistral released a roughly 1-trillion-parameter model it describes as the most capable open model developed outside China. It is launching in preview to developers, cybersecurity leaders and state authorities, with a wider release planned later this month.
cnbc.com ↗At Team '26 Europe, Atlassian announced the Agentic Multiplayer Protocol for shared human and agent workspaces. It also introduced Rovo Work, a mode for multi-step tasks that people review and approve.
siliconangle.com ↗Reflection AI released Beam, its first open-weight model. The company says it matches Chinese open models on reasoning benchmarks at lower compute cost, though the claims have not been independently verified.
techcrunch.com ↗TikTok launched an AI shopping assistant along with one-click in-app checkout. The move ties AI recommendations directly to purchasing.
techcrunch.com ↗OpenAI is launching an Intelligent UI for ChatGPT that adds interactive elements such as tappable buttons, charts and editable graphs, chosen by the chatbot as needed. It rolls out with GPT-6 to Pro, Plus, Business and Enterprise users first, with free and Go tiers following.
techcrunch.com ↗SiliconANGLE covers the GPT-6 rollout and the accompanying Intelligent UI in ChatGPT. Responses can now include interactive visual components rather than plain text.
siliconangle.com ↗Google opened its SynthID Detector, a web tool that identifies images, video and audio made with models from Google, OpenAI, Nvidia and Kakao. Sign-in is required, and it cannot localize edits or detect content from many other models.
siliconangle.com ↗