A Bitcoin trader just got $400,000 back from a wallet he'd written off more than a decade ago, and he did it with Claude.
The trader had lost access to his wallet in 2013 and couldn't recover it for 11 years. According to Tom's Hardware, he worked with Claude to build a script that systematically tested password variations based on his own memory of how he used to construct passwords back then - things like capitalization patterns, character substitutions, and the specific words and numbers he favored in 2013. The script ran through 3.5 trillion combinations before hitting the correct one.
This isn't Claude doing anything exotic. The AI's job was essentially to interview the wallet owner about his past password habits and translate those memories into systematic rules that a brute-force script could execute. The owner supplied the fragments of memory; Claude structured them into something a machine could actually run against the encrypted wallet file.
The story has a practical implication for anyone sitting on locked crypto: AI assistants are genuinely useful for this kind of password archaeology. Not because they can guess passwords, but because they help you externalize and organize your own memory into testable patterns. If you remember a word, a year, a character you always added, Claude can help you build a methodical approach around those fragments rather than just throwing random combinations at the problem.
There's also a subtler point here. The fact that 3.5 trillion attempts were needed suggests the search space was still large even after Claude's help narrowing it down. Modern hardware can test billions of passwords per second, so the bottleneck wasn't compute - it was figuring out which 3.5 trillion to try out of the near-infinite universe of possible passwords. That's where having an AI help you reconstruct your 2013 self actually mattered.