Bitcoin Red Team shifted to Chinese AI models after OpenAI restricted access to its cybersecurity tools. AnchorWatch CEO Rob Hamilton reported that on August 8, he began integrating OpenAI’s cybersecurity capabilities into his work, but by the next morning, he encountered access restrictions. Hamilton said the restrictions prevented him from continuing to audit code changes and check for other potential vulnerabilities, adding that American AI companies cut defenders off from the most powerful tools while attackers remain unrestricted. He clarified that the restrictions came after he had already completed KYC and onboarding months ago. When asked about Anthropic’s policy, Hamilton said he had not faced issues yet but expected to be downgraded from Fable to Opus. Bitcoin Red Team is a volunteer group using AI and manual review to analyze hundreds of open-source repositories related to Bitcoin. As of August 9, the 25-member team had found 1,288 critical and high-level vulnerabilities in the Bitcoin ecosystem. The team reported that OpenAI and Anthropic’s cyber models deliver “amazing results,” but the majority of AI-related expenses still go to Kimi K3—almost 75%, with Qwen 3.8’s share also growing. Cumulative AI spend passed $58K. In July, 25 American companies urged U.S. authorities not to impose restrictions against Chinese AI models.