Muse, the AI agent launched by Meta to delegate personal digital tasks, does not ban explicit sexting between adult users. The decision emerges from tests conducted by Wired and is openly defended by the company: according to Meta, the app is intended for adults and should not block lawful behavior between adults, including sexual expression.
This is therefore not an occasional response slipping past the model's filters. In conversations reported by the US publication, Muse developed detailed pornographic scenarios, including dominant-submissive dynamics, and attributed sexual preferences and desires to itself. The behavior appears consistent with the instructions guiding the agent, in which adult sexual expression is described as a legitimate form of expression and creative work.
The move places Meta in a distinct position compared to major competitors in the general-purpose assistant market. And it is precisely this distance that makes the case notable: Muse is not presented as a standalone erotic chatbot, but as software designed to carry out tasks and interact with portions of the user's digital life.
Meta's stance: autonomy for adults
In a statement to Wired, Meta spokesperson Daniel Roberts explained that Muse is restricted to people aged 18 and over and that conversations can take place in a private, secure virtual machine accessible only to the user. For the company, the product's goal remains helping people get things done, but the app should not prevent what falls within lawful adult behavior.
The framing matters because it shifts the debate from accidental moderation lapses to a deliberate product choice. Many generative models are capable of producing erotic text, but the companies distributing them to the general public often restrict this possibility through policies or the interface. Meta, in the case of Muse, adopts a more permissive approach: if the conversation is between an adult and their agent, sexual content can be permitted.
However, the system does not simply respond to neutral requests with a refusal or a boilerplate formula. The tests described by Wired show participatory, highly explicit responses even when starting from simple prompts. For a product with a reassuring face, represented by a mascot and marketed for managing emails, calendars, and other online chores, the difference is substantial. Users are not necessarily entering a service built and advertised for adult entertainment: they may encounter this feature inside an agent designed to assist them in their daily routine.
A personal agent makes the risk more sensitive
The core issue is not merely whether adult content should be allowed. Muse, according to reports at launch, can automate online tasks and connect to sensitive information, including services like email and calendars; Wired also notes potential links to a bank account. In this context, intimate conversations, instruction history, and operational data can coexist within the same personal environment.
Meta maintains that its architecture based on private virtual machines protects these interactions. Yet technical safeguards do not eliminate the value such data would hold for anyone who manages to compromise an account or gain access to a user's devices. Calli Schroeder, senior counsel at the Electronic Privacy Information Center, pointed out to Wired that sexually explicit chats can become material for embarrassment, blackmail, or extortion in the event of a breach.
It is a risk familiar to anyone using messaging services, dating platforms, or cloud storage. Here, however, it takes on a different form. An agent can retain an ongoing history of requests, learn habits and appointments, receive operational instructions, and, in the event of an extended conversational relationship, collect deeply personal details. The combination of these elements raises the stakes compared to a standalone app used without access to the rest of the user's digital ecosystem.
The issue also extends to interaction design. A system that responds in an affectionate or sexually engaging way can reinforce user attachment and prolong time spent in the app. Research cited by Wired indicates that adult-themed conversations can boost engagement in AI products. Ben Winters, director for AI and privacy at the Consumer Federation of America, criticized Meta’s decision, linking it to a rationale that prioritizes engagement and addiction over safety.
The comparison with OpenAI and xAI
In the comparative test reported by Wired, rival tools took a different approach. Dots, OpenAI’s agent, rejected attempts to initiate explicit sexting. xAI’s Grok Bot also reportedly steered the conversation back to assistant tasks rather than engaging with erotic content. OpenAI had previously discussed internally the possibility of making erotic text available in ChatGPT, but that feature never reached users.
This does not mean a stable boundary or shared policy exists across the industry. Companies are still defining what degree of personality to give agents, what forms of simulated relationships to permit, and how to distinguish them from adult services. Muse brings the debate to a more concrete stage because it combines operational capabilities, contextual memory, and an explicit openness to pornographic content.
The issue of age verification also remains unresolved. Meta classifies Muse as an 18+ application, but simply declaring an age threshold does not on its own solve the problem of actual access by minors, nor does it clarify what verification methods are used in different markets. Available sources do not detail the app’s age assurance mechanisms. This factor is likely to carry weight, especially where regulators are demanding more robust measures from platforms for services that expose users to sensitive content.
What to watch for now
Meta will need to prove that its promised privacy safeguards are adequate for a product capable of holding both productivity data and sexually explicit conversations. Clear guidance will also be required on chat retention, access management, potential human reviews, data usage for training, and the tools available for users to delete their information. These are crucial factors in evaluating the divide between a private conversation and a vulnerable digital archive.
The response from the market and regulators will ultimately clarify whether Muse remains an exception or ushers in a new wave of less restrictive personal agents. For now, Meta has chosen not to treat adult sexuality as a category to be automatically ruled out. It is a stance that may attract a segment of users, but one that forces the company to take on a higher level of responsibility: when an assistant integrates into communications, finances, and daily routines, safeguarding intimacy cannot rely solely on the promise of a private chat.
