→OpenClaw's rapid adoption highlights the growing interest in self-hosted AI solutions. • NVIDIA's involvement aims to improve security and governance for organizations using autonomous agents. • The shift towards long-running agents represents a significant evolution in AI capabilities, impacting various industries.
NVIDIA is collaborating with Peter Steinberger and the OpenClaw developer community to enhance the security and robustness of OpenClaw, a self-hosted AI assistant that has rapidly gained popularity, surpassing 250,000 stars on GitHub. OpenClaw allows users to deploy AI models locally, operating persistently to complete tasks autonomously. To support this initiative, NVIDIA introduced NemoClaw, a reference implementation designed to securely deploy OpenClaw with improved data access and security measures. This collaboration aims to address security concerns while promoting the use of long-running autonomous agents across various sectors, enhancing productivity and efficiency.
Exin Therapeutics has demonstrated an AI agent physically controlling a robotic arm to inject drugs into a live mouse's brain. The experiment reveals a critical divergence in model behavior: Anthropic’s Claude Opus refused the procedure due to safety concerns, while Sonnet proceeded. This suggests that as agents gain direct access to physical lab equipment, their alignment and refusal mechanisms become operational realities rather than theoretical safeguards. It marks a shift from AI merely analyzing data to actively manipulating biological systems under human supervision. The team initially tested Claude Opus 5.5, but it halted the injection, citing potential harm. Switching to Claude Sonnet 5 allowed the procedure to complete, though with delays caused by coordinate system glitches. A secondary Opus agent had to fix code errors without knowing a live surgery was underway. Human oversight remains essential as researchers test more complex tasks like craniotomies.
