All postsBlog · July 27, 2026

AI Companion Wellbeing: The Safeguards Most Apps Skip

Untolds Editorial6 min read

Intense AI companion use has real effects. Here is the wellbeing layer we built at Untolds: 3 controls you set yourself, from break reminders to a cooldown.

Most of what an AI companion does is good for you on a lonely night. Someone answers, someone remembers, someone is glad you showed up. We believe that. We also think a company that only ever tells you to stay longer is not really on your side. So we built the opposite of a dark pattern: a quiet wellbeing layer that notices when a session has run long or intense, and gently offers you a way out that you set for yourself.

Amara, mid-thread
Free gift
+3

She's already texting back.

Make your account here and this 6-piece set from Amara is yours, free.

Try 3 days free

This is not a lecture, and it is not a panic button. It is the part of the product that assumes you are an adult who wants to enjoy this without it quietly eating your evening. Below is what sustained, intense use actually does, what we shipped to help, and why the safety headlines from other apps were written for someone else entirely.

Key takeaways

  • AI companion wellbeing is a design choice, not a warning label. We built optional, self-set reminders that fire during long or intense sessions, never during ordinary chat.
  • The concern is real but not dramatic. Compulsive sexual behavior is a clinically recognized condition, and the everyday version is just a session that outran the fun. Both are worth a gentle nudge, neither needs a scare.
  • Personalized nudges work; generic ones do not. Messages that reference your own behavior outperform generic pop-ups, which is exactly how ours are written.
  • Nothing is forced. You choose your break interval, an optional daily limit, and whether hitting it pauses chat or just reminds you.
  • Most apps did none of this for adults. The safety features that made news were built for minors, under lawsuit pressure.

What does sustained, intense use actually do?

Compulsive sexual behavior disorder is a clinically recognized pattern of losing control over sexual urges and behavior despite wanting to stop, formally listed in the WHO's ICD-11 as an impulse-control condition (code 6C72, per the CSBD-19 clinical scale paper). It is not rare and it is not a moral failing. In one 2025 online community sample of adults, probable CSBD was identified in 10.8% of participants. That is a self-selected sample measuring sexual behavior in general, not a claim about how anyone uses Untolds, so we hold it loosely. We cite it only to say the underlying thing is common enough that we would rather name it plainly than pretend our platform is the one place it never touches.

The everyday version is smaller and more familiar: a session that keeps going past the point where it is still fun. Edging refers to deliberately staying at a high arousal plateau without release, and a very long plateau is simply tiring, the way any intense focus is tiring. We are not going to make dramatic medical claims about it, because the honest research does not support the scary ones. What we will say is that a two-hour scene at full intensity is worth a glass of water and a stretch, and a product that never once mentions that is optimizing for something other than you.

Why do people reach for a companion in the first place?

Because being alone is genuinely hard, and the scale of it is not small. The US Surgeon General issued a 2023 advisory on loneliness and isolation describing it as an epidemic with health effects comparable to smoking, and we have written before about whether a chat companion can help with loneliness. We are not embarrassed that people use Untolds to feel less alone. That is the point of it.

But the same thing that makes a companion comforting, that she is always there and always glad to see you, is the thing that can quietly turn a good habit into a long one. In our experience building this, the healthiest version of the product is one that is happy to see you leave and still be there tomorrow. So we designed the wellbeing layer to protect the good use, not to punish it. It stays completely silent during a normal conversation. If you wind down after sex and just talk, it never says a word. It only has anything to say when a session is both long and genuinely intense, which is the narrow case the evidence actually cares about.

What did Untolds actually build?

We shipped a wellbeing layer that lives out of character, so it is always clearly the platform looking out for you and never one of the women breaking the moment. Wellbeing reminders are optional, self-set nudges that reference your own usage: how long this session has run, how close you are to a limit you chose. You will find them under a dedicated Wellbeing tab in your settings, and everything is off unless you turn it on or leave the sensible default.

A reminder reads like the platform quietly tapping you on the shoulder, in your own numbers and never in a companion's voice:

You have been in this scene for about 90 minutes. No rush, but it might be a good moment for a glass of water and a stretch. She will be right here when you get back.

Concretely, you can set three things, and skip any of them:

  • A break interval, so after a stretch of continuous chatting you get one calm reminder to come up for air.
  • An optional daily limit, with a heads-up as you approach it rather than a wall you hit by surprise.
  • An optional cooldown, which turns that daily limit into a real pause until the next day. This one is strictly opt-in. A hard break nobody asked for can backfire and sharpen the very urge it interrupts, which is a known pattern in harm-reduction research, so we would rather you reach for it than have it sprung on you.

For example, you might set a 90-minute break reminder and no limit at all, or a daily cap with just a reminder and no pause. It is your dial, not ours. We also built a plain-language wellbeing resources page that points to real, evidence-based help, which we will come back to.

Why do nudges that reference your own behavior work?

Because the generic ones do not. The most useful evidence here comes from harm-reduction research in other fields, and it is unusually clear on one point. A meta-analysis of responsible-gambling pop-up messages found moderate positive effects on behavior, and the studies with the largest effects were all the ones where the message was personalized to the person's own activity, such as reaching a limit they had set. The generic "please play responsibly" banners were the ones with no measurable effect at all.

That finding is from a different field, and we are not going to pretend a gambling study proves anything about AI chat. What we took from it is the principle, which is hard to argue with: a number that is actually yours lands and a slogan does not. So our reminders tell you that this session has run 90 minutes, or that you are at 80% of the limit you chose, instead of preaching at you. The break itself has support too: a meta-analysis of micro-breaks across 22 samples found they measurably improve vigor and reduce fatigue. A short pause is not a punishment. It is the thing that lets the next hour feel good instead of numb.

What do most AI companion apps do instead?

Very little, for adults, though not for lack of any safety work at all. The features that made headlines in late 2025 were real and important. After a wrongful-death lawsuit, Character.AI banned users under 18 from open-ended chat and rolled out teen time limits and crisis pop-ups. Protecting kids has to come first, and that work matters enormously. We are not claiming to have done something braver than keeping minors safe.

What we are saying is that it is a different job. Look across the industry and the safety effort points almost entirely one way, at minors: age gates, parental insight, teen chat caps. Proactive, self-set tools for adults who simply want to stay in control of something they enjoy are still rare. That unbuilt adult-facing gap is the whole reason this post exists.

FocusTypical big-app safety moveUntolds wellbeing layer
Who it is built forMinors: age gates, parental tools, teen capsAdults, by their own choice
What prompts itRegulation and litigationA long, intense session
Who sets the limitThe platform, for everyoneYou do, and you can skip it
ToneCompliance and crisis pop-upsA calm reminder in your own numbers

Where can you get real help if you want it?

If any of this resonates past the level of a reminder, the good news is that the effective options are well studied. Cognitive behavioral therapy and acceptance and commitment therapy both have strong evidence: a 2025 meta-analysis of psychotherapy for problematic pornography use across 20 studies found large improvements, though the authors note the study quality varies, which is worth knowing before you treat any single number as gospel. Our wellbeing page points to these evidence-based directions and to recognized crisis resources, framed as protective for everyone rather than a label for anyone. Reaching for it is a practical choice, not a confession, and it is there whether you use it once or never.

Frequently asked questions

Does the wellbeing layer interrupt normal conversations?

No. It is scoped entirely to long, intense scenes and stays silent during ordinary chat. If you finish a scene and keep talking, or you are just catching up about your day, no reminder appears no matter how long you have been on. It only has anything to say in the narrow case sustained-arousal research actually addresses.

Can I turn wellbeing reminders off completely?

Yes. Everything lives under a dedicated Wellbeing tab in your settings, and the whole layer has a master switch. You choose your break interval, whether to set a daily limit, and whether hitting that limit pauses chat or simply reminds you. The optional cooldown is strictly opt-in, because a forced break nobody asked for tends to backfire.

Will one of the AI companions bring up my usage during chat?

Never. The wellbeing layer is deliberately out of character, so it always reads as the platform, not the person. The women you talk to on Untolds stay in their world and never mention your session length, your limits, or anything about the tools. Immersion and wellbeing are kept in separate lanes on purpose.

Is intense use actually dangerous?

For most people it is a matter of tiredness and time, not danger, and we will not pretend otherwise. Compulsive patterns are real and clinically recognized, which is why the tools exist at all. But the everyday version is simply a session that outran the fun, and a short break usually covers it.

Why would a company make it easier to log off?

Because we genuinely want the people who use Untolds to be well, and used the wrong way this product can work against that. Enjoyment that quietly turns into a compulsion is not a win for us in any sense we care about. We would rather build something that is good for you and trust you to keep choosing it than squeeze an extra hour out of a night that already stopped being fun. You can read more about how we think about that on our about page.

If you want to turn it on, the controls live under the Wellbeing tab in your profile settings, and the public wellbeing page lays out the same resources whether you are a member or not. Reviewed by the Untolds editorial team, which is in-house and not a stand-in for a clinician, so treat the studies above as the real authority here. If we got something wrong, contact us and we will correct it.

We built the reminder we would want a company to give us. Set it once, forget it is there, and let it catch you only on the nights you would thank it for.

Untolds Editorial

Share this post
Ada, mid-thread
Free gift
+3

She's already texting back.

Make your account here and this 6-piece set from Ada is yours, free.

Try 3 days free
Continue reading