Yes, agreed. But the bot could just add random alt text or post descriptions. Checking if the text makes sense involves using an LLM with vision capability ($).
- 5 Followers
- 0 Following
- 1 Posts
Hi @MrKaplan. I will continue my efforts to improve my moderation bot by following all your recommendations, such as implementing persistent sessions and adhering to the bot rules of the instance. Additionally, I will slow down consecutive calls to the API. Let’s see how it goes.
Thank you.
It’s fairly easy for a bot to post at random times. However it’s possible to create a bot to check for bot behavior.
I made a few tests with bot moderation and quickly reached Lemmy.world instance limits (because the bot needs to read the posts consecutively and also get the user’s public data).
When the account reaches the limit, it can’t log in for a period. I don’t want my account to be banned, so I will not continue trying to create moderation bots for Lemmy instances.
New anti-bot rules
New anti-bot rules - Lemmy.World
Hi everyone, I think it has become clear that many bot accounts are swarming the communities. Sometimes we can’t be sure whether a specific account is a bot, but there are some clues. I will not post the clues here since they could be used by bot creators to bypass the rules. Lemmy (the software) currently doesn’t have tools to automate moderation, so everything has to be moderated manually. That said, I will be adding a few anti-bot rules to the /c/comicstrips community: * A limit of two posts per person per day. * Bots are not allowed. There will be heavy moderation of suspected bot accounts. If an account is identified as a bot, its posts will be deleted and it will receive a permanent ban. And thank you to everyone in the community who reports possible spammers and bots. It helps a lot. Here is the previous post that I fixed before, which contains a few discussions about this subject: https://lemmy.world/post/42127013 [https://lemmy.world/post/42127013] Do you have any comments or suggestions? I’m listening. (This is not the future that Asimov expected.)
/c/comicstrips is experiencing this issue as well. It’s terrible.
Good to know that the devs are doing something about it.
I am pinning this post because I think it is a global Lemmy problem, and debating this issue is necessary to find ways to decrease this type of spam.
Personally, I am reluctant to implement solutions like ‘account age verification’ or ‘maximum daily posts’ because this also affects legitimate users. However, we may need to do this if we do not find better solutions.
