Bitcoin Red Team returns to Chinese AI models after OpenAI block
The volunteer initiative Bitcoin Red Team, known for its large-scale audits of open-source code in the Bitcoin ecosystem, has once again encountered barriers from American AI providers. AnchorWatch CEO Rob Hamilton reported that on August 8, he began integrating OpenAI's cybersecurity capabilities into his workflows, but by the next morning, access was severely restricted.
Access Issues After KYC
Notably, the block occurred after fully completing the KYC procedure and finishing the onboarding process, which Hamilton had completed months earlier. In his statement, he emphasized that the restrictions prevented him from continuing to audit code changes and check for new potential vulnerabilities. This is especially concerning given that the team has already identified 1,288 critical and high-level issues across 390 Bitcoin projects.
According to the expert, such policies by American AI companies effectively disarm defenders, leaving attackers with complete freedom of action. Hamilton also noted that he has not yet encountered issues with Anthropic but expects a downgrade from the Fable model to Opus, which he says is "the same story."
Chinese Models as an Alternative
As a result, Hamilton and his colleague under the alias calle were forced to return to Chinese open-source AI models. According to the team, despite the "stunning results" of OpenAI and Anthropic's cyber models, the bulk of accumulated AI expenses (nearly 75%) still goes to Kimi K3. Meanwhile, the share of Qwen 3.8 from Alibaba is rapidly growing.
The initiative's total expenses have already exceeded $58,000, indicating the serious scale of the research being conducted. Interestingly, in July, 25 American companies urged U.S. authorities not to impose restrictions on Chinese AI models, but the situation with Bitcoin Red Team shows that the problem remains acute.
My analysis: This incident exposes a systemic issue: while regulators and corporations try to restrict access to advanced AI technologies, real cyber threats know no borders. The irony is that it is precisely the open Chinese models, which they are trying to restrict, that become a lifeline for those protecting Bitcoin's critical infrastructure. While American providers play bureaucratic games, hackers have long been using all available tools.