AI security scanners are very good at overwhelming maintainers with one-off issues where enormous walls of text describe a single `list.append` call that allows memory to grow without bound. (Ironically AI coding tools are also good at putting in these kinds of mistakes, since they are so universally pervasive in the training data.) They are not good at identifying systemic fixes that address the root cause of these issues. So, let's do this: https://github.com/twisted/twisted/issues/12941
0 Boosts
0 Likes
In addition to this general work allowing for the creation of anti-bot defenses for users, it's also the kind of work that creates anti-bot defenses for maintainers. We _desperately_ need more people to triage security reports and work on fixes for Twisted right now, and unfortunately that work requires some level of access-control grants on the github org. If you are interested in such a thing please reach out. Jia Tan if you're out there just send me a request for creds
Much as in politics, real anti-AI activism is granular, slow work like addressing issues like this, and not just yelling at people online (mea culpa). Right now, Twisted would be a *terrible* way to protect your site from scraper bots; scraper bots could easily accidentally knock it over, and a coordinated agent-swarm attack would have no problem finding new attacks to do it with absolutely trivial amounts of compute. I would like it to be an excellent tool to do so.