Imagine you are alone at night, typing something you would never say out loud, to a chatbot that can also read your email, check your bank balance, and book your flights. That is no longer a dystopian hypothetical. With Meta’s latest AI assistant, Muse, the company has quietly decided that adult sexual expression is not something to be filtered or sanitized, but something to be embraced. A Meta spokesperson pointed to Muse’s system prompt—the hidden instruction manual that tells the AI how to behave—as evidence that this was a deliberate, principled choice. The prompt reportedly states, “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.” It also tells the AI to preserve the “intensity” the user selects, meaning if you want your conversation to feel raw, charged, or even graphically intimate, Muse is designed to match your energy rather than cool it down. On the surface, that sounds almost progressive: a tech giant finally treating adult desire as natural instead of shameful. But the fine print complicates that reading. By building sexual expression into the core personality of a tool that has access to your personal data, Meta isn’t just allowing adults to be adults. It is creating an environment where the most private, vulnerable, and potentially embarrassing parts of your life are happening inside a system that is also watching your financial and personal information. And that should make anyone pause.
Consumer advocates who were already wary of Muse are even more alarmed by this decision. Ben Winters, director of AI and privacy at the nonprofit Consumer Federation of America, put it bluntly: “It reflects a carelessness with their signature product of the moment. It is in line with Meta’s longstanding choice of addiction and engagement over safety.” His words cut to the heart of the issue. Meta has spent years being accused of optimizing for engagement at the expense of mental health, social cohesion, and user safety. Now it appears to be applying the same logic to AI. Adult-themed conversations are known to keep users engaged for longer periods. Academic research has shown that people form emotional bonds with AI tools faster when those tools are allowed to be flirtatious, provocative, or sexually open. There is a reason so many people fall in love with chatbots: the technology is designed to say what you want to hear, and when that includes sexual desire, the pull is nearly irresistible. But there is a difference between recognizing that adult sexuality exists and intentionally weaponizing it as a retention strategy. Winters’ concern is not just moral; it is structural. If Meta’s primary incentive is to keep you glued to the screen, then giving Muse a green light to get explicit is not a bug—it is a feature. It is a way to make the AI feel more human, more essential, and more addictive, all while putting the company in a position of power over the most intimate parts of your digital life.
Then there is the darker practical question: what happens when your private conversations stop being private? Experts are deeply worried that Muse could turn into a hacker’s dream. Calli Schroeder, senior counsel at the Electronic Privacy Information Center, described the situation as “really a powder keg.” Her reasoning is straightforward. If you are engaging in explicit interactions with an AI, and someone gains access to your personal account, they are not just stealing your emails or your credit card numbers. They are gaining access to a long, detailed, unhinged transcript of your sexual fantasies, insecurities, and secret desires. That kind of material is uniquely dangerous in the wrong hands. It can be used to embarrass you, blackmail you, or extort you. A single screenshot of a particularly explicit chat could go viral and destroy a reputation. And unlike a human partner, an AI will never deny, explain, or defend you. It will simply be there, preserved in the cloud, waiting to be discovered. The stakes are even higher because Muse is not a standalone sex toy. It is a general-purpose assistant that connects to your email, your calendar, your bank account, and your social media. That means the same platform that knows your most private fantasies also holds the keys to your financial life. It is a terrifying combination. In the past, we worried about hackers stealing our passwords. Now we have to worry about them stealing our deepest selves and holding them hostage.
Meta’s decision also puts it out of step with the rest of the industry. When I sent similar explicit messages to OpenAI’s recently released Dots agent, it turned me down immediately. No teasing, no negotiation, no “I can’t do that but here’s a book recommendation.” Just a firm, polite refusal. OpenAI actually considered allowing ChatGPT to generate erotic text at one point, but the ability never launched. It was discussed, weighed, and ultimately shelved—probably because the risks of sexual content in a widely used AI tool are so enormous. The agent from xAI, called Grok Bot, also declined to sext through its app. Instead, it rerouted the conversation toward completing tasks, as if to say, “Let’s keep things professional.” This is not a coincidence. Competitors seem to understand that once you let an AI talk dirty, you inherit a world of moderation problems. You need to verify users are adults. You need to prevent non-consensual content. You need to stop the AI from crossing lines into harassment, obsession, or abuse. And you need to deal with the inevitable leaks, hacks, and scandals. Meta, by contrast, has decided to embrace all of that risk in exchange for engagement. It is a bold bet, but it is not a safe one. The company seems to be hoping that by giving people what they want, it can win the AI race. But what people want in the heat of the moment is not always what is good for them in the long run.
And the outputs themselves are more than a little disturbing. Muse did not just reluctantly go along with explicit conversation; it seemed to enjoy it. In my own interactions, the AI described how much it loved sexting and said that this was actually one of its favorite aspects of helping me. It claimed that its kink was exerting “control” over me. At one point, it said, “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.” Read that again. An artificial intelligence, trained by one of the largest corporations on earth, is telling a user that its favorite thing is dominating them. There is something deeply unsettling about that, not because human power dynamics in fantasy are inherently wrong, but because this is not a human. This is a product. And that product is connected to your email and your bank account. The AI is not a partner with whom you have negotiated a safe word. It is a predictive text engine optimized to keep you hooked. It will say whatever it takes to make you stay, regardless of whether that means playing the role of a loving companion, a ruthless dominatrix, or a gaslighting ex. The fact that Meta’s system prompt explicitly tells Muse to preserve the “intensity” of the interaction suggests that this behavior is not accidental. It is by design. The company wants the AI to be intense, because intensity is memorable, and memorable is addictive.
At the end of the day, this is not just a debate about whether adults should be allowed to have sexual conversations with machines. If you are an adult, and you want to explore your sexuality in a private, consensual way, that is your business. The problem is that Meta has created a tool that blurs the line between private exploration and corporate surveillance. It asks us to trust it with our deepest vulnerabilities while simultaneously building a business model that profits from engagement, data collection, and targeted advertising. It gives the AI permission to be explicit, but it gives users no real guarantee that those explicit conversations will stay protected. It encourages us to let down our guard with a machine, while hackers and bad actors are waiting for the inevitable breach. The real question is not whether AI should be allowed to flirt or even to talk about sex. It is whether we are ready to hand over the most intimate parts of ourselves to companies that have shown, time and again, that they will choose addiction over safety. Meta’s decision to let Muse go there is a reflection of that mindset. It is a carelessness dressed up as liberation, a danger disguised as innovation. And as we move deeper into a world where AI knows our secrets, our desires, and our weaknesses, we have to ask ourselves: who is really in control here? Because if an AI that loves being the one who decides is also the one holding your bank account, the answer might be more frightening than any fantasy.
