AI can find zero-days but still can’t reliably write secure code

In recent months, LLMs have gone from flooding open-source projects and bug bounty programs with questionable security reports that wasted developers’ time, ...