What could've prevented the Codeberg incident?
safelyx.com
safelyx.com
However, I've been interested in the subject, and I can see how we'll need more moderation, especially, with AI content flooding everything, as well as, completely autonomous AI agents interacting with content and people.
Is this kind of "safety" check really the main process to prevent mass emails going out? Otherwise is it a free for all as far as the ability to mass email out?
There are many other ways to prevent or mitigate this specific kind of situation, as were suggested in Codeberg (Forgejo is technically the software it builds upon), like limiting email notifications to a number of mentions/people, adding a "report" button in the comments, sending notifications in smaller batches, but in reality, while these limit the attack surface area, it'll still go out to someone initially, whereas with content moderation, it's much less likely it'll go out to _anyone_. I see those as _complementary_ to automated content moderation.
I mean, I get that people received unexpected notifications and hate speech. But how? Did an admin send them? Did a bot go on all the projects? Can't we summarise that in 2 lines before going into a whole article about how to prevent "it"?
You might've heard about someone accidentally sending email to thousands of people on Github a couple years ago: https://github.com/EpicGames/Signup/pull/24