Today: MLLMs Fail to Refuse when Using Tools Agentically; Agentic AI with Structured CoT for Enhancing AI's…; Agentic Cognitive Depth: Operational Criteria for…

Today: Can an Open Model Do Security Research? Cantina’s…; After Siemens Re-Licenses OpenRadioss, Rocky Linux…; DeReAct: Decomposed Reasoning and Acting for Reliable…

Today: doxx.net Raises $38 Million to Prevent AI…; OpenAI safety leader quits, warning AI company’s…; DeepSeek Harness v0.2 Brings Official Desktop Apps to…

Today: IBM Brings Bob to Self-Hosted and Air-Gapped…; Meta open sources code to let you make Muse AI gadgets; AI chatbots remove hijabs from images of Muslim women…

Today: Nvidia launches Open Agent Safety Platform to…; AWS Strands Labs Releases Strands Decider 2B: An Open…; Cohere Releases Embed 5: How It Compares to Voyage 4…

Today: OpenAI Releases GPT-6.1 Sol: Near-Astra Coding and…; AIM: Agentic Idea Management for Automated Research; Google Releases New Gemini Model With Guardrails Amid…

Today: Reco Raises $55 Million for Agentic Security; OpenAI announces ‘dots’ agent after scrapping launch…; How Condé Nast built multimodal video discovery with…

Today: Working with AI: A Design Framework for Human-AI…; SMARtCARE: Privacy-Preserving Agentic AI Systems for…; Recorded Future Launches MCP, the Intelligence Layer…

Today: BioEVAL: A global, multi-institutional benchmark of…; ScopeBench: Do Agents Preserve Engagement Boundaries…; 2026 in LLMs (so far)

Today: China and US Agree to Establish AI Safety Channel and…; Oxford lets OpenAI train its AI models on Bodleian…; On Anthropic’s AI Misuse Report