Catenaa, Saturday, August 15, 2026-A volunteer Bitcoin security initiative says artificial intelligence has helped uncover more than a dozen vulnerabilities after scanning about 150 repositories, highlighting how rapidly AI is changing both sides of crypto cybersecurity.
The project, described as a Bitcoin “red team,” uses AI systems to examine software from an attacker’s perspective and identify weaknesses before malicious hackers can exploit them.
AnchorWatch CEO Rob Hamilton said the initiative has spent about $20,000 across AI services while developing the platform.
The team is using several frontier AI models, including OpenAI’s GPT Sol, Anthropic’s Claude models, Kimi K3 and Z.ai’s GLM 5.2. The systems help identify potential vulnerabilities and prepare technical documentation for developers.
The researchers have not identified the affected Bitcoin projects or disclosed technical details of the vulnerabilities, limiting independent assessment of their severity.
However, pseudonymous Bitcoin developer Calle said the project has reported critical vulnerabilities affecting several projects and is testing wallets, cryptographic libraries and other Bitcoin infrastructure.
The broader development could mark an important change in crypto’s cybersecurity race.
Traditional software auditing depends heavily on human researchers reviewing large amounts of code. AI systems can examine repositories continuously and compare patterns across enormous codebases, potentially finding obscure weaknesses much faster.
But attackers can use the same capability.
That risk has already surfaced elsewhere in crypto. Researchers using AI earlier uncovered a years-old vulnerability in Zcash that could potentially have enabled counterfeit ZEC creation.
Coinkite has also said it believes AI helped attackers discover the vulnerability behind the recent Coldcard Bitcoin wallet exploit.
Bitcoin bridge Boltz temporarily suspended its swap service after reporting that attackers were identifying vulnerabilities faster than developers could fix them.
The emerging problem is therefore not simply whether AI can discover crypto vulnerabilities. It is whether defenders can find and patch them before attackers do.
The Bitcoin red-team project is developing an open-source AI auditing platform that could eventually make advanced security testing available to more developers.
That could turn AI-powered auditing into a routine part of Bitcoin software development.
Yet the same technology lowers the time and expertise required to search for exploitable code.
For Bitcoin and the wider crypto industry, cybersecurity may increasingly become a machine-speed contest in which both attackers and defenders are asking AI the same question: Where is the weakness?
