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
Qwen3.5-27B 8-bit vs 16-bit Performance Comparison
A Reddit user tested Qwen3.5-27B with vLLM comparing bf16 weights and 16-bit KV cache against Qwen's fp8 quantization with 8-bit KV cache, finding practically identical results on the Aider benchmark using an RTX 6000 Pro.
Gemma 4 Early Signals: Deployment Fit Over Hype for Local Agent Workflows
Gemma 4's launch emphasizes deployment across hardware tiers with official positioning for personal hardware and edge/mobile, NVIDIA's NVFP4 quantization showing 4x compression with 99.7% baseline retention on GPQA, and Arena rankings placing the 31B dense model around #27.
Goldman Sachs Analysis Shows Minimal AI Impact on 2025 US GDP Growth
Goldman Sachs economists report AI investment contributed 'basically zero' to US GDP growth in 2025, citing imported hardware and unmeasured productivity impacts as key factors.
GPU Power Consumption Deviates from Token Predictor Theory in Small LLMs
An experiment testing the 'stochastic parrot' theory on four 8B-parameter models found GPU power consumption often scales non-linearly with token count, with divergence rates ranging from 7.7% to 36.7%. The study also revealed persistent residual heat after philosophical queries and order-dependent effects.