Money for a report and a patch, with convincing test cases, might be worthwhile. Even if a machine generates them.
Why? If it is a purely machine generated report there is no need to have dozens of third parties that throw them around blindly. A project could run it internally without having to deal with the kind of complications third parties introduce, like duplicates, copy paste errors or nonsensical assertions that they deserve money for unrelated bugfixes.
A purely machine generated report without any meaningfull contribution by the submitter seems to be the first thing you would want to exclude from a bug bounty program.
The problem they had before was a financial incentive to sending reports, leading to crap reports that wasted time to review. Incentivizing sending reports + patches has the same failure mode, but they now have to waste even more time to review the larger quantity of input.
Anyway, for most cases I'd bet that Daniel can produce and get reviewed a correct patch for a given security bug quicker than the curl team can review a third-party patch for the same, especially if it's "correct, but ai-written".
If I find a security issue, I'm willing to responsibly disclose it, but if you make me pay, I don't think I will bother.
Punishing bad behavior to disincentivize it seems more sensible.
Hence the threat to shame publicly I suppose.
Actually, Daniel Stenberg previously responded to this proposal the same way as me [1] (and maybe would still do). Coincidentally, I was reading your answer at about the same time as this part of the talk.
[1] https://www.youtube.com/watch?v=6n2eDcRjSsk&t=1823s (via https://news.ycombinator.com/item?id=46717556#46717822)
It would discourage drive-by reports by people who just happened to notice a bug and want to let the maintainers know, but I think for a project that's high-profile enough to be flooded by bogus bug reports, bugs that random users just happen to notice will probably also get found by professional bug hunters at some point.
I wouldn't do the above, but it is easy to see how I could run that scam.
Not saying I have a better solution, just that it's a hard problem. Maybe dissuading some good people who have genuine security issues but don't feel like paying just has to be a cost of doing business.
That sounds wonderfully meritocratic, but in the real world, a machine generating it is a very strong signal that it's bullshit, and the people are flooding maintainers using the machines. Maintainers don't have infinite time.
Throw enough AI crap at enough projects and you may get a payout.
The incentives fail in the face of no-effort flooding. They accidentally encourage it.