Claude AI Recovers 11-Year-Old Bitcoin Wallet Worth $400K by Finding Backup and Fixing Brute-Force Bug
A Bitcoin holder who lost their wallet password while 'stoned' 11 years ago recovered 5 BTC (~$400K) using Anthropic's Claude AI. According to X user cprkrn, they had been trying to brute-force their way in with btcrecover but failed. After finding an old mnemonic seed phrase in a college notebook that matched an encrypted wallet file, they dumped their entire college computer files into Claude as a last resort.
Claude discovered an older backup wallet file from December 2019 hidden in the data. The AI also identified a bug: the shared key and passwords that btcrecover was trying weren't being combined properly. Once the bug was fixed and the older wallet (predating the password change) was used, Claude ran btcrecover successfully and decrypted the private keys. The user then transferred the 5 BTC to a current wallet.
Early cryptocurrency wallets often mixed HD key trees (from seed phrases) with non-HD and imported keys stored in password-protected wallet files. This user's wallet contained specific keys that could not be recovered by seed phrase alone, requiring the password. The AI's ability to parse unstructured files and identify relevant backups and configuration issues proved critical.
📖 Read the full source: HN AI Agents
👀 See Also
Microsoft Keeps AI Capex Unchanged at $190B — First Hyperscaler to Hold the Line
Microsoft kept its $190B AI capex forecast unchanged, bucking the hyperscaler trend of ever-rising spending. Stock surged 8%. Meanwhile, memory chip prices may explain 45% of capex growth industry-wide.
Anthropic's Platform Strategy and the OpenClaw Response
A developer analyzes Anthropic's recent restrictions on external Claude integrations as a deliberate platform strategy, arguing for building portable stacks rather than relying on provider goodwill.
Meta Releases BOxCrete AI Model for Concrete Mix Design
Meta has released Bayesian Optimization for Concrete (BOxCrete), an open-source AI model for designing sustainable concrete mixes using U.S.-produced materials. The model improves on previous versions with better noise robustness and slump prediction capabilities.
Visual Reasoning Benchmark Results for 15 Multimodal AI Models
AIMultiple benchmarked 15 leading multimodal AI models on 200 visual reasoning questions across two tracks: chart understanding and visual logic. Gemini-3.1-pro-preview and Gemini-3-pro-preview lead the overall results, followed by GPT-5.2, Kimi-K2.5, and GPT-5.2-pro.