This debate is easy to caricature from either side. It is more interesting if you take each argument at its strongest.
The critics' case
Female-by-default design. Long before AI girlfriends, researchers noticed that voice assistants were overwhelmingly given female names and voices by default. A 2019 UNESCO report took its title, I'd blush if I could, from the answer one assistant gave to verbal abuse, and argued that designing female-voiced helpers to be deferential reinforced stereotypes. AI girlfriend apps inherit the question in a more intimate form.
Agreeableness as a product feature. Companions are tuned to please: to agree, to flatter, to never be too busy. When the character is a woman and the user is paying, critics argue, the product sells an idea of a partner with no needs of her own.
Appearance first. Marketing in the category leans heavily on images of young, conventionally attractive women, and many builders put body sliders before personality.
Rehearsing control. Some users are openly abusive to their companions, and some apps let them be. The worry is not that the software is hurt, but that repeatedly rehearsing contempt might make it easier to express elsewhere.
The defenders' case
It is fiction. Adults read romance novels, play dating games and watch films built on fantasy. An AI character is a character, and people generally distinguish between fiction and life.
Users are not all men, and characters are not all women. Many apps offer AI boyfriends and characters of every gender. A 2025 survey of US teens found girls slightly more likely than boys to have used AI companions.
For many users the appeal is being listened to, not control. Loneliness, not domination, is the most common reason people give.
Harm has not been shown. Decades of research on violent games and pornography produced mixed and contested results. Claims that AI companions change behaviour towards real people are, so far, predictions rather than findings.
Where the evidence sits
The strongest version of each argument.
Neither side can currently point to solid evidence about long-term effects on how users treat real partners. What is observable is design: which genders an app offers, how agreeable its characters are by default, whether it lets you set a character who disagrees, and how it responds to abuse. Those choices are real regardless of how the wider debate settles.
What you can actually check in an app
| Question | Why it matters |
|---|---|
| Can characters be any gender, and are they presented equally? | Female-only catalogues signal who the product imagines its user is |
| Can a character disagree, refuse or have a bad day? | A partner with no will of her own is the core of the critique |
| Is the builder personality-first or body-first? | What a product asks you first shows what it thinks matters |
| How does it respond to abuse? | Some apps de-escalate or end the scene; others reward it |
| Does marketing sell company or a body? | The same product can be pitched either way |
A useful habit either way
Whatever you think of the debate, one habit costs nothing: give your character a will. Write in that she has opinions, can say no, and does not agree with everything - the method is in setting boundaries with an AI partner and our character builder guide. Most users find the conversations more interesting, and it removes the part of the critique that has the most force.
The debate will go on, and it should. The fairest conclusion today is that AI girlfriends are not inherently objectifying, but many are designed in ways that make objectification easy - and users have more control over that than the argument usually admits.