@atoponce I get CVEs without AI, despite my best efforts.
If you create new code, some proportion of it will have security problems, even purely manually. What's the point of pretending only AI causes CVEs, it's obviously false.
We manage our manual failures by static analysis and fuzzing to try to find the flaws first. That will also work with flaws introduced by AI, right?
Services like Jules are becoming workable. I'm afraid we are all going to have to adapt to that.