Community says Claude found $86M BTC hack flaw in 8 minutes

A Reddit user in a Bitcoin-focused community claimed Anthropic’s AI model Claude identified the Coldcard code vulnerability tied to the theft of about 1,367 BTC in roughly eight minutes. Using the ID Impressive-Gene-421, the user said in the subreddit that Claude Code was simply asked to check for vulnerabilities and, after about eight minutes of reasoning, identified the critical flaw behind the hack. The user added it was hard to believe someone using AI could steal about $100 million worth of BTC simply because no one had properly reviewed the source code. Meanwhile, the attacker behind the Coldcard vulnerability had carried out a fourth attack targeting hardware wallet users as of this morning, bringing cumulative losses to 1,367 BTC, or about $86 million.