The upcoming AI safety challenge: model-to-model communication issues

In May 2025, AI researchers at Palisade Research ran a controlled experiment in which several advanced AI models, including OpenAI’s o3, were placed inside command-line sandboxes. The test was designed to measure controllability. Most models complied: Claude, Gemini, and Grok models allowed shutdown

Related Stories

Whistleblower takes legal action against xAI, alleges wrongful termination over safety issues

📰

Cloud Project Hits Major Milestones with v0.6 Release

Ex-xAI engineer alleges dismissal after voicing Grok safety worries

Mythos Safeguards by Anthropic Raise Concerns Over Lasting Inequality