You may have heard about locked accounts with bitcoins, or a web series about a man who discovered the lost password of a crypto account containing bitcoins. What if I told you that this occurred in real life? A user claimed that Anthropic’s AI chatbot Claude assisted in gaining access to a long-dormant Bitcoin wallet containing 5 Bitcoins, which had reportedly been inaccessible for more than eleven years.
The claim surfaced via a post on X by a pseudonymous user identified as cprkm, who said he had been locked out of the wallet since around 2015 after changing the password and later forgetting it. As per publicly visible blockchain activity, the wallet had shown no movement for years before suddenly becoming active again on May 13.
In a series of posts, the user explained how he discovered an old mnemonic recovery phrase linked to the wallet. However, the phrase alone was reportedly insufficient because the wallet password had been changed following the initial setup.
In a final attempt, the user stated that he uploaded files from an old college computer to Claude. The AI assistant allegedly scanned through historical wallet backups and discovered an older wallet file that could still be decrypted with the recovered mnemonic phrase. That process reportedly resulted in the recovery of the same Bitcoin private keys that controlled the funds.
The user then stated that while the encryption password for the wallet had changed, the private keys that were related to Bitcoin holdings were not changed. After successfully unlocking an older encrypted backup, access to funds was restored.
This got a great deal of attention online. Millions of views sparked the discussions about AI assisted digital forensics, password recovery and cybersecurity. Many users praised the incident as an example of how advanced AI systems can help with reverse engineering and recovery tasks that would otherwise take humans a long time to complete.
At the same time, the viral thread sparked questions about the future of encryption and cybersecurity. Some users questioned whether AI tools would eventually disprove presumptions about password-based security and hidden implementation details.