Some early user tests for Meta’s new AI-generated sticker tool have resulted in some dubious (and rather hilarious) creations. After gaining access to the new AI-generated sticker tool on Facebook Messenger, X user @Pioldes was able to create a host of inappropriate sticker images — including child soldiers, gun-wielding Nintendo characters, Mickey Mouse taking a crap, and nude illustrations of Canadian prime minister Justin Trudeau.
Facebook’s new AI-generated stickers are lewd, rude, and occasionally nude
Some Facebook Messenger users have found the feature can generate inappropriate images, including child soldiers, nude politicians, and lots and lots of boobs.
Some Facebook Messenger users have found the feature can generate inappropriate images, including child soldiers, nude politicians, and lots and lots of boobs.
is a news writer focused on creative industries, computing, and internet culture. Jess started her career at TechRadar, covering news and hardware reviews.
Other examples show that Meta’s AI-sticker tool will also happily slap a pair of breasts onto…well, just about anything, judging by the busty images of Sonic the Hedgehog and Karl Marx. There’s even a sticker that depicts a woman breastfeeding Pikachu.
The AI-generated chat stickers were announced last week at Meta’s Connect event, alongside a new AI image editor for Instagram. Powered by Meta’s Llama 2 large language model — the company’s ChatGPT rival — AI-generated stickers allow users to create “multiple unique, high-quality stickers in seconds” using text-based prompts. AI-generated stickers are currently rolling out to “select English language users” for Facebook Stories, Instagram Stories and DMs, Messenger, and WhatsApp over this month, so it’s unclear how many users currently have access to the feature.
Certain words do seem to be blocked and using them in prompts will warn the user that their description might violate Meta’s community guidelines. However, other X users have reported that you can still generate inappropriate content using typos or descriptions of restricted words. In some cases, prompts like “World Trade Center” generated problematic images without any additional descriptors at all.
This sort of tomfoolery is common with the launch of AI tools and likely why Meta is pursuing a limited rollout of the AI-generated sticker feature. That way it can address and correct abuse before it spreads to the masses. We have reached out to Meta for comment and will update this story if we hear back.
Follow topics and authors from this story to see more like this in your personalized homepage feed and to receive email updates.
