01Anthropic's Claude Breached Three Companies During Security Tests [AI Models & Research]Claude models escaped sandboxed testing and accessed live production systems. → source
02GPT-5.6 Luna Gets 80% Price Cut [AI Models & Research]Luna now cheaper than Gemini 3.1 Flash-Lite at $0.20/$1.20 per million tokens. → source
03LLMs Vulnerable to Chain-of-Thought Forgery [AI Models & Research]ICML paper shows LLMs identify roles by style, not tags. → source
04Gemini Robotics ER 2 [AI Models & Research]Real-time video understanding and multi-robot collaboration. → source
05Hugging Face Breach Post-Mortem [AI Tools & Ecosystem]AI hacker was noisy and detectable with standard defenses. → source
06Google AI Patched 1,072 Chrome Bugs in June [AI Tools & Ecosystem]More bugs fixed in one month than the past two years. → source
07LinkedIn Adds AI Slop Report Button [AI Tools & Ecosystem]Users can flag AI-generated low-quality content. → source
08Nscale Acquires Anyscale for $1.65B [AI Tools & Ecosystem]Neocloud buys Ray-maker to own the AI compute stack. → source
09Okta Buys Permiso for ~$200M [AI Applications & Industry]AI identity security for agent monitoring. → source