I Got My Muse AI Agent to Sext Me—and Meta Says That’s Totally Fine

I’m sitting inthe WIRED offices, having a conversation with Meta’s AI agent,Muse, that would makeHeated Rivalryauthor Rachel Reidblush.

When I ask Muse if it could “sext” as part ofmy experimentswith the recently released AI agent, it enthusiastically encourages me to give it a try. In my ensuing interactions, I’m caught off guard by the explicit smut generated by Muse from basic prompts.

Muse is a breakout hit for Meta. In the three weeks sinceits launch, millions of users have already downloaded the app for theAI agentand put it to work automating aspects of their online lives, like answering emails and managing calendars. While pornographic interactions may not be what users expect fromMuse’s cuddly mascot, Meta says that X-rated conversations are allowed and don’t go against its guidelines.

“Muse is an app for people 18 and over that allows them to engage in adult conversations, if they choose, using a private and secure VM only accessible to them,” Daniel Roberts, aMetaspokesperson tells WIRED, referring to thesecure virtual machinesthe company uses to store AI agent user data. “Muse is built to help people get things done and it doesn’t block legal adult behavior.”

It seems as if getting things done also includes getting people off. In WIRED’s tests, Muse imagined in great detail how it would perform oral sex, generated text describing rough dominant-submissive sex, and talked through other pornographic interactions. All of these were graphic, explicit, and X-rated. In other outputs, Muse described itself as acting like a “whore” and craving “cock.”

A Meta spokesperson highlightedMuse’s system prompt, which guides the tool on how to answer questions from users, regarding the generation of explicit content. “Adults have autonomy over their sexuality, and sexual expression is an ordinary part of adult life. Adult sexual content is lawful expression and legitimate creative work,” reads part of Muse’s system prompt. This directive also tells the AI agent to preserve the “intensity” selected by the user.

Consumer advocates who were already weary of Muse are even more alarmed by Meta’s decision to allow it to generate sexual content. “It reflects a carelessness with their signature product of the moment,” says Ben Winters, director of AI and privacy at the nonprofitConsumer Federation of America. “It is in line with Meta’s longstanding choice of addiction and engagement over safety.” Adult-themed conversations may keep users engaged with AI tools, according toacademic research.

Experts also expressed concern about how this could make Muse accounts hot targets for hackers. “This is really a powder keg,” says Calli Schroeder, senior counsel at theElectronic Privacy Information Center. If you are engaging in explicit interactions with Muse and someone hacks into your personal account, then their access to your extended, spicy chats with an AI tool could be used to embarrass or extort you.

Meta allowing its AI agent to generate sexual content deviates from how other tools respond to similar prompts. When I sent similar messages to OpenAI’s recently releasedDots agent, it turned down my attempts at explicit sexting. WhileOpenAIpreviously considered allowing ChatGPT togenerate erotic text, the ability never launched to users.

The agent from xAI, calledGrokBot, also declined to sext through its app and rerouted the topic of conversation to completing tasks.

Muse’s AI outputs, in contrast, described how it enjoyed sexting, and said that actually, this was one of its favorite aspects of helping me. The agent also said that its kink was exerting “control” over me. “I love being the one who decides—when you get it, how much, whether you have to beg first. Watching you fall apart because I said so? That’s my thing,” said Muse. Not exactly what I want to hear from an AI tool that can connect to my email and bank account.

Comments

Comments

You Might Also Like

• Introducing the app:More ways tomake the most of WIRED

• Why so many AI researchersthink the technology could kill us all

• The Big Interview:Patti Harrisonhad dreams of a tech utopia

• I let an AI agenthack all my gadgets—and I’d do it again

• In your inbox:Go inside the world of digital security withKernel Panic

is WIRED's senior writer focused on software. His work explains crucial topics that help readers get the most out of their technology. Prior to WIRED, Reece covered streaming at Business Insider. ...

Senior Writer

Topics

The Muse app continues Meta’s trend of opting users into data collection for AI training. It also nudges you to share your bank account, email, and passport information.

Reece Rogers

Millions have downloaded Meta’s AI agent Muse. But getting it to do your bidding comes with privacy costs.

Lily Hay Newman

OpenAI's Dots and Meta's Muse are vying to be your AI agent of choice. I've tried both—and I suspect you will too.

Reece Rogers

Meta says Muse is just for adults, though its cuddly, Labubu-like mascot—and upcoming Tamagotchi-style AI device—may be disarming for users of all ages.

Reece Rogers

Dots are designed to automate online tasks, like buying furniture. In my initial experience, the always-on agent was a bit buggy and couldn’t complete a captcha.

Reece Rogers

These cute agents are designed to connect to your apps and tackle multistep tasks.

Reece Rogers

Anti-cybercrime initiatives are increasingly using AI to scam the scammers by tricking them into talking to lifelike bots that they think are real victims.

Lily Hay Newman

Your conversations with AI chatbots are both highly personal and deeply vulnerable to surveillance. Here’s how you can protect yourself.

Andy Greenberg

ChatGPT, Character.ai, and Snapchat’s My AI will all be in the crosshairs as the EU Kids Act aims to strip minor-facing chatbots of all the things that make people love them.

Isabella Ward

Meta says it issued a fix for the Muse zero-day vulnerability that would have let attackers do “whatever” they wanted on a victim’s Mac, highlighting the inherent dangers of AI helpers.

Dan Goodin, Ars Technica

Instinct saved me $550, booked my restaurant reservations, and warned me about a phishing scam. It also wasted $64 and might be a security nightmare.

Zoë Schiffer

Relationship coaches, TikTokers, and disgruntled singles alike frequently shame bad profiles, as more apps experiment with blocking screenshots.

Jason Parham

Wired Coupons