public models

VidocSecurity replicated Anthropic's Mythos research findings using publicly available models, demonstrating that similar results can be achieved without proprietary systems. The experiment highlights potential security implications for AI model transparency and safety.