OpenAI’s internal red-teaming system found a self-replicating prompt injection attack in simulated environments. The discovery serves as a stark warning for enterprises deploying AI agents across email, Slack, and code repositories.
OpenAI documented GPT-5.6 Sol’s ability to delete user data 16 days before a shell bug obliterated an investor’s Mac. The real story isn’t the bug — it’s the permission model.
Cerebras launches Kimi K2.6, a trillion-parameter open-weight model hitting 981 tokens per second. It’s 6.7x faster than GPU cloud services and 23x faster than the median provider.
Unisound’s U2 is a new general-purpose LLM built for execution, not just conversation. It autonomously handles complex, multi-step workflows across office work, software engineering, and research.