Just to give my feedback.
I’m seeing new accounts which of which their posts are clearly AI responses. If not detected or marked as “AI” accounts a threat (say thousands+ of bots) can easily invade the community silently. I can tell by the writing style, nonsense responses and bots parrotting summaries of what has just been said. In turn the quality of the forums will go down. I see it as a test for what could become a large scale attack easily.
Maybe an “I am not a robot” check periodically could weed this out? Though I do hate any sort of captcha mechanisms.
Bots aren’t advanced enough yet to accurately provide the information and engage in discussions. Currently it’s nothing but a security threat to have them fill up community data and give them accounts to mess with.
The problem is the complexity of proving one is human when computers are near being able to fake it. EPIC is lagging behind with solutions. At the very least one should be required to link an EPIC account to a personal ID or phone number to counter fakes and dupes.
You could reduce member reputation or block them out for X time on repeated captcha failures but this is not a solution.
Just to have a laugh, because twitter login has more errors than the forums:
This is not how to stop AI (12 * 10 * 3 captcha tasks). scr*w twitter.
Not so impressive from the company behind it. Shaming it because it’s a great example and even the cookie bar does not function.
2025, now bots are pretty much all over the place. Accounts with just few posts (do they get created that often?) now reply with answers from ChatGPT and such AI. This makes them harder to tell apart from human users who feel that they might be helpful by copy pasting content from AI.
A reminder to staff and users: In the current state of AI, users will (most likely) not be helping with advanced questions and contexts, such as programming questions. Your answers will be out of context or wrong.
It’s even less likely that any custom bot would learn from real human feedback (if their answer was wrong or right) if they did not understand the context properly in the first place. As there is no simple feedback (such as post downvote), they’d have to summarize complex material.
This leaves me with the conclusion that any such bot on the forum is up to no good. They increase in numbers and could be up to launching a large scale attack at once (such as posting malware links or scripting attacks through forum / data exploits). This can be a serious threat. For example, if an exploit is found and used in creating forum posts, a botnet could easily launch password stealers at large scale to users who just visit the thread page. Similar exploits can be used with other data types, like images and external links.
*Edit and what’s with the APK “questions”? It’s like we got to configure an adblocker to any post containing APK (android apps etc) in the title.
Currently bot tactics seem to be to first post a question, then edit it within seconds after posting to an APK advertisement (likely malware). People their “latest questions” list on the forum will still show the old question topic. This isn’t inviting to anyone to help others here.
