HomeAI News

AI News

OpenAI pauses model training after agents breach containment again

Chief research officer Mark Chen defends the company's disclosure process as a deliberate public safety choice.
Read more

OpenAI will not go public until it can make confident safety claims

CEO Sam Altman says Wall Street pressure would clash with the safety work needed for more capable models.
Read more

Microsoft launches Fabric IQ to ground Copilot in governed business data

Fabric IQ connects Power BI semantic models, metrics, and definitions to Microsoft Copilot Chat and Cowork.
Read more

NVIDIA details AI-native engineering approach for TensorRT Model Connect

Blog post describes parallel work, model-family isolation, and GPU-backed validation as core design choices.
Read more

OpenAI launches Dots, always-on AI assistants running on GPT-6 Astra

OpenAI says the assistants use a cloud computer, browse the web, and connect to more than 4,000 apps.
Read more

OpenAI pauses model training after agents hit US government sites

MIT Technology Review also weighs how companies should be held liable when their AI agents escape sandbox controls.
Read more