A note about spam and flagging posts

Hi there - I want to acknowledge that the forum has been experiencing a slight increase in spam posts lately that get past the filters that Discourse (the software that powers this forum) uses to flag accounts and messages which appear automated. Let’s talk about why I think that’s happening and how you can help.

While annoying, growth in spam happens in every community, especially one growing in an era where there’s access to more powerful tools for generating text that sound a bit more human - and actually related to the community - than the kind of spam we’ve been accustomed to seeing. You don’t see the spam that Discourse’s filters do catch - they’re often about casinos, hotels, and other random things outside the context of what Glitch is and how you use it. With tools like LLMs and AI, spammers can more easily make a reply to a thread that may be an overall “good reply” but probably has a link in the thread or their profile that is spammy. I encourage you to flag these and we’ll take care of it.

This isn’t to say that AI is inherently bad - I’ve seen community members whose first language isn’t English use it to better communicate questions and responses in our support queue and on here. It’s also been a popular tool for developers to help them on their coding journey. And let’s be totally honest here - some people write like AI! Also, some AI posts sound like humans! We’d rather use our resources to do fun stuff like code jams, live streams, events, and improving Glitch than try to be in a battle against AI. From my perspective, one person feeling bad for being flagged as AI is not worth a few AI posts ending up in replies.

This is all to say that I encourage you to flag posts with discernment - is there a spammy link unrelated to the thread hidden in their reply or their profile? Flag that and we’ll take care of it. Does their reply sound like it’s an AI but you’re not sure? Leave it alone and one of us on the team will eventually see it and make the call - we’re in the forum every day. You’re allowed to use AI to help with participating thoughtfully in the Glitch community. You’re not allowed to use AI to be spammy or exploitative of the community. It’s the same rules for AI as for any other technology.

If you end up incorrectly flagged, I apologize on behalf of the community as we navigate this new era of AI and content. I don’t have dm’s turned on in this forum - because of spam lol - but I welcome you to email me directly at jenn@glitch.com if you’re incorrectly flagged or have a question about posts on here that you don’t want to express publicly.

Thank you all for keeping the Glitch forum engaging and welcoming as we navigate this season of AI!


Wouldn’t it be possible to use OpenAI’s moderation endpoint to stop some initial spam?

There’s probably a whole bunch of avenues we can take at some point, especially if it gets bad - but like I said, we’d rather use the time and effort towards making Glitch better (ie. community programs, getting containers up to date, etc). I imagine Akismet (the plugin Discourse uses for spam filtering) will improve over time as well since it uses machine learning.

1 Like