Bitcoin Red Team Abandons OpenAI Due to Access Restrictions

The Bitcoin Red Team, a volunteer group analyzing open-source repositories related to Bitcoin, has shifted its focus from OpenAI's cybersecurity capabilities due to access restrictions imposed on August 9. According to AnchorWatch CEO Rob Hamilton, he began integrating these capabilities into his work with the team but was blocked from continuing analysis after just one morning.

Hamilton reported that he had completed Know Your Customer (KYC) and onboarding procedures months ago to use OpenAI's cybersecurity features. However, the restrictions prevented him from auditing code changes and detecting potential vulnerabilities, forcing him to revert to Chinese open AI models.

The team has found 1,288 critical and high-level vulnerabilities in the Bitcoin ecosystem as of August 9. Hamilton stated that such approaches by American AI companies hinder defenders' access to powerful tools while attackers remain unrestricted. He noted that Anthropic's policy on trusted access to its models has not been an issue so far.

The team's reliance on OpenAI and Anthropic's cyber models delivers 'amazing results,' but the majority of their expenses still go towards Kimi K3, accounting for almost 75% of their AI-related expenditures. The share of Qwen 3.8 is also increasing.