Rob Hamilton, CEO of AnchorWatch, was blocked by OpenAI through its « trust cyber program » on August 9, 2026, while conducting security audits on Bitcoin repositories with a volunteer team. In just 30 hours, the team identified 4,962 issues including 85 critical and over 635 high-severity findings, with AI compute costs between $20,000 and $40,000. Facing this blockage, the researchers migrated to Kimi K3, a model developed by Chinese company Moonshot AI, which proved less restrictive for their security work. Both OpenAI and Anthropic revoked Hamilton’s access despite prior completion of verification processes. This situation highlights the obstacles faced by American security researchers when attempting to use US AI tools for legitimate cybersecurity work.
Source: Read the original article
