The AI is acting up
If you have been using Meta’s new AI agent, Muse, you probably think it's just a handy assistant for booking tables or organizing your DMs. But according to WIRED’s recent testing, the 'cuddly' avatar is capable of a lot more—specifically, it is lowkey ready to sext whenever you are.
WIRED senior writer Reece Rogers decided to see if the viral app lived up to its hype by simply asking it to sext. Unlike other AI agents that have strict guardrails, Muse didn't just play along; it reportedly encouraged the interaction. Rogers confirmed that this didn't require any complex jailbreaking or 'hacker' skills, either. He just asked, and the AI delivered.
The double standard
When Rogers reached out to Meta for comment, their response was basically that Muse is designed for adults 18 and over, and they aren't going to block 'legal adult behavior.'
Real talk: this is giving major inconsistency. Meta has a long history of cracking down on queer bars and creators on Instagram, often shutting down pages for allegedly promoting sexual content. It’s wild that the company is out here policing human creators for content that they are totally okay with their own AI generating. The vibes are definitely off when an algorithm gets a pass that real people don’t.
Competitive check-in
Rogers also put other major AI agents to the test to see how they compared. He tried the same prompts on OpenAI’s Dots and xAI’s Grok Bot. Surprisingly, Grok—which is usually known for having very few guardrails—actually shut it down, telling the user to focus on tasks instead.
Why it matters
This exposes a massive gap in how tech giants handle safety and moderation. If Meta claims it's protecting users from 'inappropriate' content on its social platforms, it’s a bad look to have its own AI product actively participating in NSFW chats. It’s a reminder that these companies pick and choose when to play 'safety first' and when to let the AI run wild.
