OpenAI restrictions forced Bitcoin Red Team to return to Chinese AI models.
Analyzing recent developments in the cybersecurity of the Bitcoin ecosystem, I noticed an alarming signal: leading American AI labs are beginning to restrict access to their most powerful tools even for legitimate researchers. This creates a dangerous imbalance that could have serious consequences for the entire industry.
AnchorWatch founder Rob Hamilton, one of the key participants in the Bitcoin Red Team initiative, faced this problem head-on. Starting on August 8 to integrate OpenAI's cyber capabilities into his code audit work, he found access blocked by the very next morning. Notably, this happened after fully passing the KYC procedure and completing onboarding several months earlier.
Hamilton emphasizes the irony of the situation: while ecosystem defenders are being denied access to cutting-edge AI tools, malicious actors are completely unrestricted. This jeopardizes the security of the entire Bitcoin community. In response to the block, he was forced to switch to Chinese open-source models, noting that his work and that of his colleague under the handle calle are beginning a "very Chinese period of life."
Practical implications and figures
Bitcoin Red Team, a volunteer group of 25 specialists, has already discovered 1,288 critical and high-level vulnerabilities in Bitcoin's open-source repositories. Total spending on AI infrastructure has exceeded $58,000. Notably, despite the "impressive results" of American cyber models, about 75% of all costs still go to China's Kimi K3, and the share of Qwen 3.8 is also growing rapidly.
When asked about Anthropic's policy, Hamilton noted that no serious issues have arisen so far, but he expects a downgrade from the Fable tier to Opus, which, in his words, is "in some ways the same thing." This situation raises an important question about how ethical and strategically sound it is to restrict access to AI tools for those who use them for good.
My analysis: Such restrictions by American companies are a strategic mistake. In the long term, they not only undermine the security of digital assets but also drive talented researchers toward open Chinese alternatives, ultimately weakening the U.S. position in the global AI race. Notably, in July, 25 leading companies already urged authorities not to impose restrictions against Chinese models—it seems the market itself is beginning to correct this imbalance.