AI tools raise security concerns in blockchain and healthcare

Security experts warn that artificial intelligence is making it easier for hackers to exploit flaws in older smart contracts, threatening the billions of dollars locked in decentralized finance platforms. Continuous monitoring, rather than one‑off audits, is now recommended. A recent AI‑driven audit of the privacy‑focused Zcash blockchain uncovered a four‑year‑old vulnerability, and analysts estimate that hackers have already stolen about $1.3 billion from smart contracts this year. Similar attacks on the defunct Aztec Connect and the mySwap exchange illustrate the ongoing risk.

A separate study of European physicians shows that, despite a new AI regulation requiring clinicians to detect and correct AI errors, doctors rarely do so. In experiments with fabricated patient cases, doctors followed AI classifications even when outcome data plainly contradicted the AI’s advice, leading to biased treatment decisions and the illusion that ineffective therapies were beneficial. The findings highlight automation bias and the need for better oversight of AI‑assisted medical decisions.