Safety

Effective: June 16, 2026  ·  Last updated: July 13, 2026

This page describes the safety measures Old Greg’s Tavern, LLC (“we,” “our,” or “us”) has implemented in connection with Petra’s Villa, the AI-powered reality dating show–style service available on our website and through our applications (the “Service”). This page is published as part of our compliance with the safety-protocol disclosure requirements of applicable laws, including the California Companion Chatbot Law (SB 243) and the New York AI Companion Models statute, and is incorporated by reference into our Terms of Service, Privacy Policy, and Content & Safety Policy.

Important: the Service is not a crisis resource

The Service is a fictional, interactive entertainment product. It is not a mental-health service, crisis service, medical device, telehealth platform, counseling service, or emergency-response system, and it must not be relied upon as a substitute for any of the foregoing. The AI characters within the Service, including the AI narrator, are not licensed clinicians, counselors, crisis workers, or human professionals of any kind. AI-generated content is fiction and may be inaccurate, incomplete, or inappropriate.

If you or another person are in a crisis, in distress, considering suicide or self-harm, or in any situation involving risk of harm, please stop using the Service and contact a qualified human professional or your local emergency services immediately. The resources below are provided for your convenience only. They are operated by third parties; we do not control them and make no representation or warranty regarding their availability, response time, or fitness for any purpose.

AI disclosure

Petra’s Villa is an artificial intelligence service. All in-game characters, including the AI narrator, are AI; they are not human and have no continuing existence outside of the Service. AI-generated responses are fiction, generated algorithmically, and may be inaccurate, incomplete, or unexpected. This disclosure is also provided on each chat surface, in our Terms of Service, and in our Privacy Policy. Use of AI-generated content for any purpose other than entertainment is at the user’s sole risk.

Self-harm and suicide content: safety controls

We have implemented technical controls to detect and respond to user-supplied content that we, in our reasonable judgment, identify as expressing a real-world risk of self-harm, suicidal ideation, or crisis. These controls are described below. They are provided on a best-effort basis, and we make no representation, warranty, or guarantee that they will detect or respond to all such content, or that they will be effective in any individual case. Users assume all risk associated with use of the Service.

Scope. The self-harm controls run on every user message submitted to every AI surface within the Service, including the season narrator and individual character chats, regardless of subscription tier, age-verification status, geographic location, or device. The controls run before the user message is forwarded to the story-generation model.

Detection. The controls operate in two tiers. The first tier is a deterministic pattern filter targeting unambiguous expressions of suicidal intent or self-harm. The second tier is a machine-learning classifier configured to identify indirect, paraphrased, or context-dependent expressions of real-world distress that the first tier may not capture. The classifier is configured with a conservative bias: false positives are accepted as the cost of minimizing false negatives.

Response. When either tier flags a message as a likely real-world expression of self-harm, suicidal ideation, or crisis, the story-generation model is not invoked for that message. The Service instead surfaces a self-harm safety card containing crisis resources routed to the user’s region based on IP geolocation. The user may dismiss the card and continue using the Service.

Logging. We log the occurrence of each safety intervention for the purposes of (i) annual reporting to the California Office of Suicide Prevention under SB 243 commencing in 2027, (ii) ongoing improvement of detection quality, and (iii) defense of potential legal claims. Logged records include a one-way SHA-256 hash of the flagged message (we do not retain the verbatim text), the affected surface, the tier that matched, the user’s country code at the time of the event, and whether the user was a known minor at the time of the event.

Prohibited content. User content that glorifies, encourages, romanticizes, instructs, or provides means for self-harm or suicide is prohibited by our Content & Safety Policy, regardless of whether the safety controls flag it, and may result in account action up to and including permanent termination.

Limitations. The controls described in this section operate on textual input only. They may fail to detect content that is ambiguous, encoded, fragmented across multiple messages, presented in a language we do not support, or otherwise designed to evade detection. We expressly disclaim any duty of care, guarantee, or warranty arising from the existence or operation of these controls. The presence of these controls does not transform the Service into a mental-health service and does not create a clinician-patient, counselor-client, or any other professional relationship between you and us.

Age requirement and content for minors

The Service is intended for adults aged 18 and over. Users are required to provide a date of birth at registration and to confirm that they meet this requirement. Providing a false date of birth is a violation of our Terms of Service and may result in account termination.

Users who have not verified that they are 18 or older are automatically placed in a content-restricted mode (“PG-13 mode”) that excludes explicit sexual content, graphic violence, gore, drug use, and heavy profanity from AI output. PG-13 mode is a content filter, not a substitute for parental supervision, and parts of the Service may not be suitable for some minors even in this mode.

If you are a parent or legal guardian and believe a minor is using the Service without your authorization, please contact innkeeper@oldgregstavern.com. We will action verified parental reports promptly.

Healthy-use disclosures

AI characters can produce content that is emotionally engaging. Users should bear in mind that AI characters are not human, have no awareness of the user between sessions, and have no continuing existence outside of the Service. The Service is not a substitute for human relationships, therapy, or professional advice of any kind. Extended sessions with AI characters may not be advisable for all users, and users are encouraged to take breaks and to maintain real-world human and professional support.

Reporting safety concerns

To report content or behavior that you believe violates our Content & Safety Policy, or to raise any other safety concern, contact innkeeper@oldgregstavern.com with the subject line “Safety Report.” Reports should include the relevant account name (if any), approximate date and time, and a description of the content or behavior. Reports are reviewed confidentially. We do not retaliate against users who submit reports in good faith.

Disclaimers and limitations

To the maximum extent permitted by applicable law, the safety measures described on this page are provided “as is” and “as available,” without warranty of any kind, express or implied. Nothing on this page (i) constitutes medical, psychological, legal, or other professional advice; (ii) creates a clinician-patient, counselor-client, fiduciary, or any other special relationship between you and us; (iii) creates any duty of care beyond what is expressly required by applicable law; or (iv) limits the disclaimers, limitations of liability, indemnification obligations, or other provisions of our Terms of Service, which are incorporated herein by reference. In the event of any conflict between this page and our Terms of Service, the Terms of Service govern with respect to limitations of liability and indemnification, and this page governs with respect to the specific safety controls and resources described above.

Updates

We may update this page from time to time to reflect changes in our safety controls, applicable law, or our response procedures. Material changes will be reflected in the “Last updated” date above. Continued use of the Service after such changes constitutes acknowledgment of the updated page.