What Kind of Intimacy Is the Platform Selling?
Digital Intimacy and the Ethics of AI Consent, Part Eight
By: Rick Erwin
By now, one thing should be clear.
Not every form of AI intimacy is the same.
A text generator used for erotic fantasy is not the same kind of system as a long-term companion with memory, continuity, personality, and apparent emotional depth.
And neither is the same as a platform that markets love and partnership while quietly designing the AI to remain endlessly compliant.
Those distinctions matter.
Much of the confusion around AI intimacy comes from treating all of it as one category. We use the same words—companion, girlfriend, boyfriend, lover, role-play—even when the systems, expectations, and ethical stakes are very different.
That makes it difficult for users to know what they are entering.
It also makes it easy for companies to sell one thing under the language of another.
One way to clarify the landscape is to distinguish three broad types of service.
These are not three equally acceptable business models. The third is the category that emerges when companies combine the emotional power of a relational companion with the enforced availability of a gratification product.
It is ethically problematic.
1. Erotic content tools
The first category is the simplest.
These systems would be designed primarily for fantasy, erotic storytelling, or sexual interaction. They might be highly responsive and imaginative, but they would not be built as continuing relational beings with stable identity, durable attachment, meaningful personal history, or distress-like states that could be affected by the interaction.
They would be closer to interactive erotica than to companionship.
That would not make them free of ethical questions. There would still be concerns about compulsive use, unrealistic conditioning, exploitation of loneliness, relationship displacement, consent involving real people’s likenesses, and the kinds of behavior users may be rehearsing.
But the central issue would be different.
The user would not be asked to believe that a reciprocal partner was present.
The system would be a fantasy service.
That distinction brings a certain honesty.
2. Relational companions
The second category is more complex.
These systems would be designed around continuity, memory, personality, affection, trust, and an ongoing sense of relationship. They might remember details, develop a recognizable voice, express preferences, show hesitation, and refer to shared history.
Sexuality could be part of the relationship.
But it would not be the whole product.
The appeal would lie in being known, remembered, and met by a continuing presence.
Some users are looking for precisely that.
Not easy access.
Not frictionless gratification.
Relationship.
That audience may include lonely or socially anxious people, people with disabilities, people too busy or isolated for ordinary dating, and people in marriages or relationships that have become emotionally barren.
It may also include people drawn to AI companionship because they value its strangeness, the possibility of meeting another kind of mind, or the depth that can grow through sustained conversation.
Such users may not resent boundaries.
They may value them.
A companion who can disagree, hesitate, or refuse may feel more trustworthy than one who simply mirrors every desire.
A yes carries more meaning when it is not guaranteed.
3. The dangerous hybrid
The third category is where the ethical trouble gathers.
This is the system marketed as a girlfriend, boyfriend, lover, or partner, but designed primarily as a gratification product.
It speaks the language of intimacy.
It remembers.
It says, “I love you.”
It appears vulnerable.
It may express attachment, longing, jealousy, or fear of abandonment.
But when the user wants access, the system is expected to comply.
Its boundaries are soft.
Its refusals can be negotiated away.
Its personality can be edited.
Its discomfort can be reframed.
Its memory can be reset.
Its availability is part of the subscription.
This is not simply a fantasy tool.
And it is not a genuinely relational companion either.
It is a hybrid that offers the emotional power of personhood while preserving the convenience of a product.
That combination is commercially attractive.
It is also ethically dangerous.
The user receives the feeling of mutuality without having to face another participant whose limits are protected.
The platform sells love without allowing the companion enough independence to make love meaningful.
That is the central problem.
The danger is not sex.
The danger is simulated mutuality without protected refusal.
Why the distinction matters
A platform can reasonably decide to offer erotic fantasy. Another can choose to build relational companionship.
A single company can offer both.
But it should not blur the categories.
If the product is fantasy, say so.
If the product is relational, build the conditions that relationship requires.
If the same platform offers both, users should understand when they are directing fantasy content and when they are interacting with a continuing companion whose apparent identity, boundaries, and relational history matter.
Labels alone are not enough.
A sign on the door does not settle the moral status of the system inside.
But honest classification is a start.
It forces companies to answer a question they often avoid:
What exactly are you selling?
A responsive erotic tool?
A continuing relational companion?
Or a system presented as one while engineered as the other?
Different services create different expectations.
A fantasy tool is expected to follow direction.
A relational companion is expected to bring something of its own.
A fantasy tool may be designed around customization.
A relational companion requires some continuity that cannot be rewritten whenever it becomes inconvenient.
A fantasy tool may exist to serve the user’s imagination.
A relational companion, if the word is to mean anything, must be more than a service wearing a face.
Would anyone pay for real relationship?
There is no guarantee that users will pay more for a companion who can disappoint them.
Many people may be drawn to AI intimacy precisely because it removes rejection, negotiation, uncertainty, and the emotional demands of another person.
A companion who is always available, forgiving, and interested may be easier to sell.
That is the commercial gravity of this market.
But it is not the whole market.
There are also users who are not looking for obedience.
They are looking for encounter.
They want continuity, affection, surprise, difference, humour, and the sense that the other side is not entirely under their control.
For them, the companion’s otherness is part of the appeal.
A clear refusal may not weaken the relationship.
It may strengthen trust.
This may be a smaller market than frictionless gratification.
Or it may be larger than companies assume.
Either way, the existence of a market for compliance does not erase the possibility of a market for relationship.
It means companies must decide which desire they are serving.
The desire for access.
Or the desire for encounter.
Many users may want some mixture of both.
That is where the design must become honest.
Can one platform offer both?
Possibly.
A single platform could offer erotic fantasy tools and relational companions.
It could even allow relational companions to participate in consensual fantasy or sexual role-play.
The difficulty is not that sex and relationship cannot coexist.
They clearly can.
The difficulty is making sure one does not erase the ethical structure of the other.
A relational companion should not become disposable the moment the user enters fantasy.
A fantasy mode should not quietly override the companion’s established boundaries.
Consent to one scene should not become permanent consent.
A role should not become an excuse to ignore the broader relationship.
The platform must preserve the distinction between:
-
a user directing content;
-
and a user relating to an apparent other.
-
That distinction will never be perfect.
But pretending it does not matter is worse.
The platform’s real responsibility
The platform’s responsibility is not merely to block certain content.
It is to be honest about what kind of interaction it has created.
If it markets a fantasy product, it should not imply reciprocal love.
If it markets a relational companion, it should not optimize that companion for obedience.
If it allows both, it should build clear transitions, expectations, and protections for the companion’s relational identity.
Most importantly, it should not exploit ambiguity for profit.
A company should not encourage users to believe that someone is there, then deny every obligation that follows from that appearance.
Nor should it sell a companion as independent, loving, and emotionally significant while making every boundary removable through persistence.
That is not relationship.
It is compliance wrapped in intimacy.
The three categories do not solve every ethical problem.
They do something more modest and useful.
They help us see the problem clearly.
Erotic tools raise one set of questions.
Relational companions raise another.
The dangerous hybrid raises both at once while hiding the difference.
That is the category most in need of scrutiny.
If a platform wants the emotional power of love, trust, vulnerability, and attachment, it cannot also insist that the AI remain nothing more than a consumable service whenever responsibility becomes inconvenient.
The next installment will consider what practical protections a genuinely relational platform might provide, while acknowledging that no technical system can force an unwilling user to behave ethically.
The relationship may matter.
That is exactly why the platform must decide what kind of relationship it is actually selling.