AI Is Rereading Crypto’s Old Code, and Finding What Humans Missed

Crypto’s security problem took a strange turn in 2026. Artificial intelligence (AI)-assisted researchers found a four-year-old Zcash flaw capable of creating counterfeit ZEC, a Coldcard weakness dating to 2021, which preceded more than $100 million in estimated bitcoin thefts, and attackers manipulated an AI agent into transferring roughly 3 billion DRB. Meanwhile, malicious instructions embedded […]

≈