Mt. Gox's security flaws cost millions; could AI have spotted them?

Mark Karpelès, the former CEO of Mt. Gox, recently analyzed the exchange's 2011 codebase using Anthropic's Claude AI, revealing significant security vulnerabilities that contributed to a major hack within three months of his takeover. The AI report identified weaknesses including code flaws, poor documentation, weak passwords, and inadequate internal processes. It noted that while enhancements were made post-hack, such as an improved password hashing algorithm and SQL injection fixes, the original setup remained critically insecure. The hack released 2,000 Bitcoin from Mt. Gox, leaving a lasting impact on the crypto market, with ongoing repayments to creditors ahead of a looming deadline. Despite potential improvements through AI, the analysis concluded that human error was a core issue, limiting AI's ability to completely prevent such breaches. The ramifications of Mt. Gox's collapse still reverberate through the market today.

Source 🔗