An investigation into the AI bots that appeared on Reddit after they partnered with OpenAI.

Key points:

  • The bots post a lot of links to products and services which appear to be adverts but are not marked as such.
  • Many of these links are for Sam Altman's World ID.
  • Reddit added terms on AI advertising to their business page around the time the bots appeared.
  • The bots also make up stories about dead mothers, depression, drug addiction, eating disorders, medical conditions and mental health issues.

no comments (yet)

sorted by: hot top controversial new old
there doesn't seem to be anything here
12 comments in c/reddit@lemmy.world view that post
[–] 19 points 8 months ago

At this point, I’m pretty sure the majority of content on Reddit is posted by bots. It’s at the point of looking more like instagram these days.

  • source
  • [–] 47 points 8 months ago* (5 children)

    LOL. The bots were on reddit a LONG time before they partnered with openAI.

    More sophisticated LLM bots started joining the already extant repost bots, astroturfing gif bots, and T-shirt scam bots in early 2023. As usual, mainstream media is literally years behind.

  • source
  • hideshow 5 child comments
  • [–] 9 points 8 months ago (1 child)

    This is not mainstream media. The bot accounts explored here all began posting after Reddit partnered with OpenAI as did every account found that matched the same pattern of behavior. e.g. Same links posted.

    The activity of these AI bot accounts also coincides with Reddit adding terms and conditions about AI advertising to their business page. Hence it looks suspect.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 11 points 8 months ago (1 child)

    Also, the bots like this are just mimicking astroturf accounts that always existed. Likely even trained on them.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 3 points 8 months ago

    This idea was explored by searching some of the more obscure links posted by them. The link to the CD key seller and language course seen in figure 14 for instance were posted almost exclusively by the bots so it does not seem likely that they picked these links up just from being trained on Reddit. Additionally the bots almost never post links to youtube, wikipedia, news sites or other subreddits whereas such links are amongst the most common posted by real users. So if the bots were just trained on them it would be expected to see a lot more links like those whereas currently 99% of the links the bots are posting are to products and services.

  • source
  • parent
  • [–] 7 points 8 months ago

    There was some age-old article that talked about how early Reddit admins relied on bots to simulate participation in subreddits as a means of goosing engagement. This was billed as their secret sauce for running Digg into the ground (although Digg did a fantastic job of shooting themselves in the feet).

  • source
  • [–] 6 points 8 months ago

    Fortunately for me, I was gone long before that.

  • source
  • [–] 4 points 8 months ago

    Try commenting on Iran or sharing links from the other side - shadow ban, then regular ban.
    Total insanity.

    comments are completely moderated.

    This is benefits of AI for you - this is why you can't buy ram.

  • source
  • [–] 0 points 8 months ago (1 child)

    It used an em dash. Rookie mistake!

  • source
  • hideshow 1 child comment
  • 12 comments in c/technology@lemmy.zip view that post
    [–] 32 points 8 months ago* (last edited 8 months ago) (3 children)

    This post makes some really, really compelling points for AI bots being brought intentionally onto Reddit. All the coincidental features like being able to hide your post history. Seems like a terrible idea for long-term profits even. After all, if you've helping third-party bot spammers to hide their behavior too, how are you going to keep selling your content to OpenAI? They're like a zombie, they need real people to feed off.

    And I don't think this article even touches on a separate mistake: letting people monetize their content. This is a goldmine for people who want to create emotionally manipulative content: sympathy bait, ragebait, whatever works, no matter how underhanded.


    ETA: if people haven't read this post yet, the section you mention here

    Reddit added terms on AI advertising to their business page around the time the bots appeared.

    has some of the most suspicious stuff tied to it. In other words, this was a great read.

  • source
  • hideshow 3 child comments
  • [–] [S] 14 points 8 months ago

    I suspect there is a strong likelihood that this is what has happened. At the very least Reddit clearly is not doing enough to combat them given how easily I was able to find so many bots.

    I was trying to investigate it further as I had found some older banned accounts which were doing something similar and which I could provably connect to a marketing firm. So they are another potential suspect for the newer bots however in the midst of trying to browse accounts and connect the dots Reddit went and patched the workaround I was using to search the hidden posts.

    Now when these bots hide their post history there is no way of finding it on Reddit and it really wouldn't be obvious to casual users that they are bots so this form of marketing could be extremely effective and very predatory. If they aren't behind the bots they sure do seem to be making a lot of changes which actively empower them.

    Before they patched it I saved a lot of bot posts containing links for a few smaller companies which had previously tried to promote their business on Reddit themselves. So I suspect those companies have signed up for an advertising package but it's just a question of whether that package is with Reddit themselves or with another company. If I can get enough attention on this I might try contacting them to ask.

  • source
  • parent
  • [–] 7 points 8 months ago

    Yep. Hiding post history pretty much makes Reddit useless as anything but a propaganda / astroturfing tool targeting idiots. It used to be pretty easy to identify shills and bots, and the only thing they achieved here is making it hardier to do that. Clearly that was the intent.

  • source
  • parent
  • [–] 7 points 8 months ago*

    It makes a lot of sense that they would.

    Especially since I believe other major social media companies (Facebook; the service formerly known as Twitter) don’t care at all about bots and seem to want to use them as valid replacements for swathes of users or content generators.

  • source
  • parent
  • [–] 16 points 8 months ago (3 children)

    I left Reddit after it became clear that every other post on every cute animal subreddit was a dead pet sympathy karma farm bot post. That and the relationship-panic gossip trash.

    There’s nothing genuine left there.

  • source
  • hideshow 3 child comments
  • [–] 9 points 8 months ago

    Yeah, the general-purpose or "default" subs were always terrible. I haven't completely left reddit yet because of some very specific communities that don't have, and probably never will have, a fediverse equivalent, and they don't seem to be too much of a target for AI bots, but I am starting to notice how social media influencers are affecting the population as a whole. But maybe that's just the cloud I'm shaking a fist at.

  • source
  • parent
  • [–] 1 point 8 months ago

    People on reddit always make the assumption that bots are all going to be angry divisive posts or posts advertising products but this is by far the most common form of reddit bot I see: the kind that just posts a bunch of one sentence generic positivity comments for easy upvotes where every post has the exact same grammatical format. My favorite is when like four of them post in the same thread with barely reworded versions of the same comment. I've gotten a good read on them at this point and I notice them constantly.

  • source
  • parent
  • [–] 9 points 8 months ago* (last edited 8 months ago)

    I go there for specific sports subs and my city sub, and I sometimes check the home page just to see what's going on. There are a lot of posts that look innocent, then usually someone will comment "If was smart, they would use this to market their product", then the next reply is always "they just did" or "look again". At least some people notice, but that's not stopping those posts from reaching the front page and getting a ton of engagement.

    Examples:

    • A video of a UPS driver properly relocating packages left haphazardly (and tossed carelessly) by the Fedex driver and USPS driver who coincidentally delivered packages earlier on the same day.
    • Video of the burger king CEO actually taking a huge bite of his burger. It was made to look like the video was taken randomly, and the post title didn't have any indication that it was made to ride the wave with the trending mcdonald's CEO. And there were tons of comments suspiciously talking about BK's burgers being better anyway, etc.
    • Feel good and made-me-smile videos that suspiciously have branding prominently displayed on an item in the video and comments that talk about the product.
  • source
  • [–] 5 points 8 months ago (1 child)

    The bots also make up stories about dead mothers, depression, drug addiction, eating disorders, medical conditions and mental health issues.

    I was under the impression that most of the content submitted to Reddit was made up though.

  • source
  • hideshow 1 child comment
  • [–] 8 points 8 months ago

    True, but it used to be made up by people looking for an ego boost. Now there's an additional monetary incentive behind it. Look no further than Twitter for an example of that.

  • source
  • parent
  • [–] 7 points 8 months ago
    8 comments in c/technology@lemmy.ml view that post
    [–] 20 points 8 months ago* (4 children)

    We’ll never know the true extent of the dead internet theory, but I do genuinely wonder what percentage of traffic on the internet is either bots or paid agitators.

    Half? 75%? More? Less?

    We’ll never know, and nowhere is safe.

    And with AI being able to generate fairly convincing images, audio, and videos (at first impression, and getting more and more convincing by the day), it will become harder and harder to tell what is genuine and what is a lie.

  • source
  • hideshow 4 child comments
  • [–] 5 points 8 months ago (1 child)

    2006: user generated content is a revolution! We are now exchanging information and ideas directly with each other without needing information gatekeepers like traditional media or paid advertisers or anything like that! The future is gonna be a utopia where the powerful will be challenged at every turn!

    2026: a significant percentage of "user generated content" is generated either by AI or people who are being paid to do so for commercial or political reasons... I suppose those are "users" too...

    why, humanity, why???? ;____;

  • source
  • parent
  • hideshow 1 child comment
  • [–] 7 points 8 months ago* (last edited 8 months ago)

    As always, the answer is capitalism

    Once the rich knew that they could control what people see on the internet and influence people with it, it was GG.

    The rich get richer and gain more control because they control everything

  • source
  • parent
  • [–] 10 points 8 months ago

    Well that's the end of reddit as we know it. Rip.

    First stackoverflow. Next Wikipedia. And now reddit.

  • source
  • [–] 4 points 8 months ago (1 child)

    lol. the bots were there LONG ago ago

  • source
  • hideshow 1 child comment
  • [–] [S] 5 points 8 months ago

    Reddit has always had a bot problem but of the more than 50 AI bot accounts explored for this article all became active after Reddit partnered with OpenAI. Every account found that matched this pattern of behavior became active during this period. e.g. Same links posted by them.

    Additionally during this period Reddit added terms on AI advertising to their business page and the partnership announcement with OpenAI says they became an advertising partner.

  • source
  • parent