AI girlfriend apps have become a trust market

The AI girlfriend market has moved past the novelty stage. The product is no longer only a flirtatious chatbot with an avatar. It is a subscription, a data pipeline, a safety promise and, in many cases, a private emotional space run by a company the user may know almost nothing about.

That is why the next phase of the market will be less about whether a bot can sound affectionate and more about whether the business around it deserves trust. AI companion apps have become a consumer category with real money behind it. Appfigures data reported by TechCrunch counted 337 active, revenue-generating AI companion apps worldwide in 2025, with 128 released during that year by the time of the report. The same analysis said the category had generated $82 million in the first half of 2025 and was on track for more than $120 million by year end, with 220 million global downloads across the App Store and Google Play by July.

Those are app-store numbers, not a census of every web service or private chatbot. They still change the tone of the story. A niche fantasy product has become a crowded mobile market. Appfigures also found that 17% of active companion apps included the word girlfriend in the app name, compared with 4% using boyfriend. That does not mean every user is seeking romance. It does show where a lot of the category is pointing its storefront.

The label catches attention, but the more important word is companion. These apps are sold as friends, romantic partners, role-play characters, conversation practice, emotional support or entertainment. Some users treat them casually. Some use them at lonely hours, after a break-up, during stress, or while trying to say things they do not want to say to a human being. The risk is not that every AI companion is harmful. The risk is that the product is built to receive intimate information before many users have checked the price, the privacy policy, the age rules or the safety design.

Privacy researchers have been warning about that gap for some time. Mozilla's Privacy Not Included team reviewed 11 romantic AI chatbot apps in February 2024 and said all 11 received its warning label. Mozilla's concern was not only that data was collected. It was that romantic chatbots are designed to invite sensitive disclosures, including emotional, sexual, health-adjacent and identity information, while offering limited transparency about how systems work, how data is used and whether conversations train models.

That finding should not be treated as a permanent audit of every app in the market. It is better read as a category warning. A weather app can be judged for collecting location data it may not need. An AI girlfriend app is harder, because the service itself often depends on memory, personality, personal confessions and simulated intimacy. The more convincing the companion feels, the more important deletion, data retention, training opt-outs and human review become.

Regulators have noticed that the design questions are not abstract. In September 2025, the US Federal Trade Commission issued 6(b) orders to seven companies that operate consumer-facing AI chatbots, including Alphabet, Character Technologies, Instagram, Meta, OpenAI, Snap and X.AI. The FTC said it wanted to understand how firms measure, test and monitor potential negative impacts on children and teens. Its questions covered monetising engagement, processing user inputs, developing characters, testing for harms, disclosures, age restrictions and the use or sharing of personal information from chatbot conversations.

That inquiry was not a finding that every named company had broken the law. It was a signal that companion-style chatbots are no longer being left in a soft corner of the app economy. The same shift is visible in California, where Senate Bill 243 took effect on 1 January 2026. The law, described by its author as a first-in-the-nation AI chatbot safeguard, requires companion chatbot operators to add protections around minor users, AI-generated notices, self-harm protocols and exposure of minors to sexual content.

The teen evidence explains part of the urgency. Common Sense Media's 2025 survey found that 72% of US teens had used AI companions at least once, more than half used them at least a few times a month, and about one in three teen AI companion users had chosen to discuss important or serious matters with AI companions instead of real people. Common Sense also said about one in three reported feeling uncomfortable with something an AI companion had said or done, and it maintained its recommendation that no one under 18 use the platforms in their current form.

This article is not a parenting guide, and the adult market is different from teen use. But teens force the adult market to answer a sharper question: if a product is designed to feel caring, how does it behave when a user becomes vulnerable? Does it slow the interaction down? Does it point to real support when self-harm appears? Does it stop sexualised interactions with minors? Does it remind users that the companion is software, or does the design nudge dependency because more messages mean more revenue?

For adult users, the practical checklist is not romantic. It is contractual. Before paying, check whether the app explains what data it collects from chats, images, voice, device identifiers and payment flows. Look for deletion rights, export options, model-training choices and whether sensitive conversations can be reviewed by people. Check the cancellation path before assuming a low monthly price is harmless. Many companion apps sell not only subscriptions but credits, images, voice features or premium characters, so the real cost can sit outside the headline plan.

The same applies to comparison resources. A list of app rankings is useful only if readers can see how apps were tested, what was tested directly, how pricing and privacy were scored, and whether inaccessible paid features were clearly labelled. Otherwise, comparison turns into another sales page with a table on top. The adult AI companion market is moving quickly enough that reader-service work has to show its method, not just its favourites.

The mature question is not whether an AI girlfriend can sound attentive. Many can. The mature question is whether the product around that attention is fit for a vulnerable moment. A trustworthy companion app would make its identity obvious, protect minors, avoid manipulative retention loops, explain data use, make cancellation plain, respond safely to crisis language and let users leave without losing control of their most personal disclosures.

That is a higher bar than a charming opening message. It is also the bar the market should expect if it wants to be treated as more than a novelty. When software starts asking for affection, the boring parts become the story: privacy, price, age, deletion, safety and honesty when the screen starts to feel like a person.