Login
You're viewing the functional.cafe public feed.
  • Nov 10, 2026, 8:25 PM

    One reason AI companies are so eager to report FOSS vulnerabilities is that doing so creates the perception of need. Doesn't matter if they're mostly low-risk or low-priority. Doesn't matter that they were mostly discovered by actual cybersecurity researchers. The perception of need is the point. Volume serves that aim. Dropping 50 reports on a project makes people to think, "Wow, we're cooked without AI." That many reports at once overwhelms the maintainers? Guess they need AI to keep up.

    💬 4🔄 99⭐ 91

Replies

  • Nov 10, 2026, 11:02 PM

    @lrhodes it's the argument that some people are using to try to force to allow LLM reports in projects with a strict no-ai policy. Some people were telling me yesterday that I'm basically killing people for ignoring LLM security reports in libxml2

    💬 4🔄 11⭐ 54
  • 💬 0🔄 1⭐ 9
  • Nov 11, 2026, 11:32 AM

    @danigm I expect this attack as well, and I plan to answer that if the issue is so damn important, and you really want to help my project, some human can take the time to verify and write up the issue. If you don't actually want to help the project, that's fine, please go away. @lrhodes

    💬 2🔄 2⭐ 14
  • Nov 11, 2026, 3:28 PM

    @bremner @danigm @lrhodes More like write down. When I get this slop I don't want a verified version of the same obnoxiously verbose writeup. I want the idea expressed ideally in one line of human language with a minimal test case in human written and human comprehensible code.

    💬 0🔄 0⭐ 7
  • 💬 0🔄 1⭐ 3
  • 💬 0🔄 0⭐ 2
  • 💬 0🔄 0⭐ 3
  • 💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • Nov 11, 2026, 6:30 PM

    @lrhodes The guy who diagnoses the problem also being the guy who sells the solution is a classic recipe for grifts. E.g. dishonest car mechanics. It's a conflict of interest that requires a lot of professionalism and integrity to overcome, and suffice it to say, AI companies have none.

    💬 0🔄 0⭐ 0