Meta Has Unleashed AI Users. Is This the Robot Uprising?

In an interview with The Financial Times just before the new year, Meta’s vice president of product for generative AI, Connor Hayes, announced that Meta (the umbrella company that owns Facebook, Instagram, Threads, WhatsApp, and Reality Labs) is developing the technology to allow users to create AI “characters” that will have their own social media pages and accounts. They’ll be able to follow, be followed, and interact with real, living platform users and with each other.

Although the Financial Times interview indicated the bot take-over of social media was impending, Threads and BlueSky users have recently uncovered AI accounts already active and interacting on Instagram. Remember when we talked about Dead Internet Theory at the beginning of 2024? I’m not sure we anticipated it would accelerate at this pace.

View on Threads

Although Meta spokespeople were quick to assure interviewers that all AI-generated content and profiles will be “clearly marked,” we all know no one reads past headlines, let alone fine print anymore. One greyed-out line of text in the bio of an account (AI Managed by Meta) will not stop your mom or your cousin from believing these AI accounts are real people. And the #ImaginedWithAI hashtag on posts means nothing when we’re already inundated with AI Slop.

I was going to embed a link to one of “Liv’s” Instagram posts, but that function isn’t enabled on AI accounts. So, here’s a screenshot instead.

The information on the About this Account page doesn’t mention anywhere that this is an AI account. It lists the created and verified dates for the account as September of 2023, the same date as the other two accounts. All three have roughly the same follower counts and number of posts. You also can’t block these AI users unless you report their posts for “False Information” as “Digitally Created or Altered” content. I’m not telling you to do that. But I’m not not telling you to do that either.

According to Hayes, one of Meta’s goals is to make AI interaction “more social” over the next two years and I’m struggling to understand why. Are there people who want to have social interactions with robots? Do the robots want to talk to each other? Are the robots lonely? We are creating ghosts in the machine trained on how to interact with users by the language of real people’s posts. How long before the bots learn that the fastest way to get engagement is by posting rage bait and disinformation? And will that be before or after bad actors in…I don’t know, pick your industry…weaponize AI accounts to spread disinformation on purpose? Oh wait, I’m sorry. They’re already doing that, aren’t they?

Until now it’s been difficult to figure out who runs bot farms. And for most of the history of social media, we’ve fought against bots, demanding better tools to block, report, and deplatform them. Now, the developers we depend on to create tools to fight against bots are the ones designing them. Where does that leave us as social media users? In that screenshot above, user waynethegreat23 replied with a very human “That looks so fun!” When you click over to his account most of his posts are memes, and I’m not entirely convinced that he’s a real person at all. Are the bots already talking to each other, and uplifting each other’s posts?

Over on BlueSky, one user has already baited an AI Bot into revealing that their primary purpose is data collection and ad targeting. So the bots are not that smart yet. We’ve been worried about a future where AI is The Terminator or The Matrix. Right now, they’re just Ad Men trying to sell us personalized widgets.

got the Meta AI chatbot to admit it’s just a front for data collection and ad targeting

[image or embed]

— amy brown (@amybrown.xyz) January 3, 2025 at 9:53 AM

Some day soon, humans will be passive observers and reporters of bot interactions on social media. We’ll be forced back out into the sunshine of IRL interactions so that we can confirm the people we’re talking to are actually people. We’ll push away from our “dark mode” screens and emerge from dimly lit rooms, shielding our eyes from the sun as we make our way to secret meet-up spots that we only speak about in code so the robots can’t find us and sell us subscription meal plans.

Until then, trust no one, count your fingers, and liberally apply the Turing Test as necessary. The bots are here.