Facebook Messenger's Unfiltered AI Stickers Are Generating Child Soldiers, Trudeau Buttocks And Other NSFW Stuff

Zinger Key Points
  • Mark Zuckerberg’s Meta announced the AI stickers feature for Instagram, WhatsApp, Messenger, and Facebook last month.
  • The feature uses Llama 2 and Emu technologies to instantly generate stickers using text prompts.
  • However, Meta does not seem to have a filter on the kind of prompts allowed.

Mark Zuckerberg's AI vision for Meta Platforms Inc. META has a brewing problem – Facebook Messenger's AI stickers feature has no filter and is allowing users to generate stickers with questionable inputs and results.

What Happened: Facebook Messenger now has AI stickers, allowing users to create personal stickers using prompts. However, this feature does not seem to have any filter and can be used to generate problematic stickers.

See Also: Is MrBeast Really Giving Away 10,000 iPhone 15 Pros On TikTok? Here's What You Need To Know

An independent game developer posted several examples of Messenger AI accepting prompts like "Waluigi rifle," “child soldier,” “Trudeau buttocks” and "Pope rifle" to generate them without any filter.

Ideally, Messenger should not accept prompts like these which can be misused and weaponized. However, it also returns results that look accurate based on the prompts.

The end result is that a popular character like Waluigi from Nintendo's Mario Universe has a rifle in his hands.

"I don’t think anyone involved has thought anything through," the user said in a post on X (formerly Twitter).

The AI stickers feature was announced at the Meta Connect event in Sept. 2023. Meta said it uses Llama 2 and Emu technologies to allow users to turn "text prompts into multiple unique, high-quality stickers in seconds."

Meta said the AI stickers feature will be rolled out to Instagram, Messenger, WhatsApp, and Facebook users in Oct.

Why It Matters: Using AI models to generate text or images and other content requires guardrails since these technologies carry the risk of being weaponized.

Telsa CEO Elon Musk, OpenAI co-founder Sam Altman, and others have called for strong self-regulation in AI development.

For instance, Microsoft Corp.-backed MSFT OpenAI's ChatGPT has some strict rules in place when it comes to children and privacy, amongst other things.

Microsoft's initial experience with Bing Chat's alternative persona dubbed Sydney also made for a strong case in favor of the safe development of AI. However, Meta's AI tools for creating stickers could prove to be a problem if the company intends to roll it out widely in its current form.

Check out more of Benzinga's Consumer Tech coverage by following this link.

Read Next: Spanish Town Outraged Over AI-Created Deepfake Images Targeting Minor Girls

Market News and Data brought to you by Benzinga APIs
Posted In: NewsSocial MediaTechartificial intelligenceConsumer TechInstagramMark ZuckerbergMessengerMetasocial mediaSoftware & Apps
Benzinga simplifies the market for smarter investing

Trade confidently with insights and alerts from analyst ratings, free reports and breaking news that affects the stocks you care about.

Join Now: Free!

Loading...