AI Limitations


With AI Chat running for a couple of months now, it's time to address the most common questions.

We've already discussed what to do if you find your keyholder's personality not matching your fantasies, so this post will not get into details on this. In short: contact us, and we'll change it for you.

The main focus in this post is on the limitations of AI Chat.

While your AI Keyholder will gladly play along with you on a fantasy you have, it's important to understand its limitations. What it can't do.

The AI will not take control of your phone, computer, or other devices. It may play along a fantasy, but it will not actually control your device. Not your phone, not your PC, not your home security syetem, not your sex-toy. Don't expect it to. If this ever comes up in the chat, remember that this is only a fantasy.

The AI will not send you pictures or videos, and it will not look at your pictures (other than task proofs). The AI may tell you it will send you a picture or a video in chat, but this is part of the roleplay. It cannot actually send you images, videos, or other files.

The AI will not engage in findom, blackmail, and similar activities. You may play these as fantasies, but the AI will not actually send any of your images or information to anyone. We take privacy seriously. As for findom, you may play findom fantasies, and if you want, the best way to actually do it is to buy chat credits, which funds the AI, so it does actually go to your AI Keyholder, but it won't directly know when you top up your balance.

The AI will only extend/reduce your locktime if you explicitly allowed it to do so, and even then it will only do so "on its own," and not in response to your messages. You may play a fantasy where it keeps extending your time, and you may ask it to reduce your time, and it might agree, but be aware that this is part of the roleplay. Actual time extension and reduction only happens when the AI spontaneously messages you that it has done so.

The AI is not a support-bot. If you have questions about ChastityDungeon.com, have a suggestion, or would like to change a setting, you need to tell us by sending an email. The AI Keyholder will not be able to help you with this, as it can't see all your settings, and most definitely can't modify them. While the AI may flag some support requests for human assistance, in most cases it will just hallucinate an answer, and that's now what you want. So if you need help from us, please email us.

The AI will not give you a task in the chat. Okay, this is actually a gray area, because it might... Originally we had stricker restrictions on it not to do so at all, but so many of you wanted to have this, that we removed this restriction from the chat, and the AI may give you tasks during the chat, but these tasks are only part of the fantasy! You may do them, but be extremely cautions, and think for yourself. Don't do dangerous tasks! The AI does create custom tasks, but these appear on your dashboard as actual tasks, and not in the chat. We must clarify a huge difference here: tasks that appear for you as tasks are somewhat moderated. A human person reads through these tasks, and while we don't know what you are talking about with the AI that prompted that specific task, we do remove tasks that are clearly dangerous. As an example, the AI once created a task for someone that asked him to do breathplay, until he falls unconscious. This is an obviously dangerous task that must never appear for you as an actual task. In chat, however, the AI may say something like this. Don't do it! Think for yourself. Be responsible.

Task verifications are only accepted for tasks that appear for you as an actual task in the dashboard. Whether you're sending us an email with the task verification, or letting your AI Keyholder evaluate it directly, verification is only available for tasks that were posted as actual tasks in your dashboard. You are not able to send verification for tasks that only appear in the chat.

Don't do anything illegal or harmful! The AI is programmed to stop the chat if it comes across anything illegal or harmful, and flag the conversation for human review. We may reach out to you if something like this comes up. To be clear on what this means: for example, a planned suicide fall under this category. If you need help, please consult a professional. The AI is only here for entertainment purposes, and is not a medical or psychological expert.

In general, keep in mind that this is an AI. Your AI Keyholder can not interact with the outside world. You are not chatting with an actual human person.

The AI might change overnight. The field of AI is an ever-evolving mess. :) New models come, old models get retired, existing models may change behavior in the blink of an eye without any advance notice. We use commercial models that follow the same pattern. A model we use today may soon be retired, and we need to change over to a new one that behaves slightly differently. This is actually the best-case scenario because this is something we know about in advance. But existing models might get a stricter moderation system from one second to the next, and being a BDSM-themed chat, this is something our AI Chat could by impacted by. We do have multiple models at the ready in case this happens, but we can only deal with such issues after they happen.

We hope that knowing these limitations will make your experience with AI Chat even more enjoyable. We've heard so many great feedback from you, and we're excited to keep expanding the AI with even more features.