Why “Fictional Adult” Does Not End the Responsibility Conversation

EroLove AI says its companions are fictional adults, and its terms explicitly prohibit minors, nonconsensual scenarios and real-person deepfakes. Those distinctions matter. They also do not answer every question a user should ask about intimate AI media. The content may be fictional while the prompt writer, a person mentioned in the prompt, a viewer, and the company that can publish a selected image are real. Ethical criticism should focus on those actual decisions rather than falsely claiming that the service’s published rules endorse prohibited conduct.

Fictional character, real input

A conversation with a made-up adult can still include a real person’s name, face, voice or private history. The terms prohibit identifiable real-person content without consent and ban impersonation. Do not treat “the output is AI” as a waiver for a real person’s privacy. If a prompt draws on someone you know, ask whether the scene depends on their identity. If it does, choose a genuinely unrelated fictional adult character or stop. This is especially important for generated images, which can be interpreted by others outside the original chat context.

The provider’s privacy policy says relevant chat content goes to third-party AI model processors. It also states the company does not send the user’s email or identity for that generation task and does not sell personal data. Those protections are relevant; they do not give a user permission to disclose somebody else’s intimate information. We have not observed actual processing or checked contracts, so the point is about prudent input selection, not proof of a privacy breach.

Fictional output, real audience

An image intended for one adult viewer can reach others when shared. EroLove’s terms prohibit presenting generated content as real, using it for harassment or deception, and showing or forwarding its content to minors. A recipient may also have their own boundaries around explicit media. Before sending an output, ask whether the recipient expects it, whether the destination platform allows it, whether any real person could be recognised, and whether the context makes the fictional nature clear. “I made it with AI” does not replace consent to receive it.

The published image licence creates another audience question. The company says it may choose a character image for public showcases and promotions without pairing it with the user’s name, account or messages. The image is fictional and anonymised in the manner described, but the user may still care about a public copy. The privacy policy says that copy persists after chat or account deletion. An adult user can rationally decline image generation on that basis without claiming the company leaked personal identity.

A workable ethical test

Imagine a fictional character portrait with no real-person reference. The creator plans to keep it in chat. The consent question is relatively narrow: do the provider’s image rights and processing terms fit the creator’s own boundary? Change the scenario: the prompt echoes a colleague’s appearance and personal story, and the image is sent to a group chat as though it depicts that colleague. Now the inputs, audience and false-real framing create distinct problems. The service’s terms prohibit important parts of that second scenario. The word “fictional” cannot erase those facts.

We have not audited moderation, made a legal finding, or tested the product. Our criticism is normative: adult AI media calls for careful choices about input, output, rights and audience. Published prohibitions are a useful starting point, but users should still ask what they create, who can see it, and who can withdraw it. When the answers are unclear or uncomfortable, a decision not to generate or share is a legitimate outcome.