A cryptocurrency investor who was locked out of his Bitcoin wallet for over a decade has successfully used Anthropic’s Claude AI to crack his own encryption, reclaiming access to a fortune worth R6.55 million ($399,404).
Read: Bandwidth Blog & Smile 90.4FM Tech Tuesday: AI trends shifting!
The user, posting under the handle @cprkrn on X, shared his success story alongside verified blockchain transaction history. The public ledger confirmed that his 5 BTC had sat entirely untouched since April 1, 2015, with the first withdrawals beginning on May 13, 2026. The entire balance has since been liquidated.
The investor admitted that his decade-long ordeal began after he altered his security credentials while under the influence. “Locked out 11+ years because I got stoned and changed the password,” he revealed, noting that the long-lost password turned out to be “lol420fuckthePOLICE!*:).”
Before turning to artificial intelligence, the user essentially staged a brute-force cyberattack against his own security files. His recovery attempts included:
- Hashcat Attempts: Utilizing Claude to help coordinate a Hashcat attack that cycled through 3.4 trillion password combinations without success.
- BTCrecover Automated Attacks: Attempting a localized brute-force attack via Claude that tested 34 billion passwords at a rate of 300,000 combinations per second, which also failed.
The breakthrough came when the user executed a last-ditch strategy: dumping the entire data contents of his old college computer into Claude’s context window. The generative AI scanned the historical files and discovered an overlooked backup wallet file containing a successfully decrypted mnemonic phrase. Expressing his gratitude, the investor publicly thanked Anthropic and its CEO, Dario Amodei, even vowing to name his future child after the tech executive.
While this specific case required local file access, meaning an external hacker could not have used Claude to breach the wallet remotely, it highlights the rapidly advancing capabilities of generative AI in code analysis and vulnerability discovery.
See also
This analytical power is increasingly being observed in the cybersecurity landscape. Earlier this month, researchers from the security firm Theori utilized a proprietary “AI hacker” named Xint Code to audit the Linux kernel’s crypto subsystem. In roughly an hour, the AI successfully scaled the audit and uncovered a high-severity vulnerability dubbed “Copy Fail”, a flaw that would have been exceptionally difficult for human auditors to isolate manually.
The ease with which AI can parse data to find security gaps is raising alarms across the financial sector. Cybersecurity experts warn that the same tools helping users recover lost keys or helping researchers patch bugs can be weaponized by bad actors to discover and exploit zero-day vulnerabilities at unprecedented speeds.
Bijan Sanii, CEO of global fraud detection provider Inetco, which secures major South African institutions like Standard Bank and African Bank, emphasized that financial entities must prepare for a new era of accelerated threats. “For institutions, the issue is that AI is accelerating the discovery, testing, and potential weaponization of software weaknesses,” Sanii warned. “AI-assisted tools can help attackers discover and exploit vulnerabilities much faster.”
