AI & Data Use
How AI is identified, how private information is processed to provide the Service, how memory works, and how optional model improvement differs from ordinary use.
Consolor is built around AI. This page explains how AI is identified, how private information is processed to provide the Service, how memory works, and how the optional model-improvement feature differs from ordinary use.
1. You Are Interacting With AI
Consolor’s Assistant and Companions are artificial intelligence systems, not humans. Consolor should identify the active AI experience clearly and should not imply that a Companion is conscious, sentient, emotionally attached, a licensed professional, or a person with human lived experience.
2. AI Is Used to Provide the Core Service
When you choose to use Consolor, AI processes the information needed to generate responses and provide features such as summaries, topic identification, recommendations, personalization, navigation, memory/context, and safety controls. This core processing is necessary for the AI service you request and is governed by the Privacy Policy, Consumer Health Data Privacy Notice, and required Sensitive / Health Data Consent.
If you upload an attachment, your Companion may analyze its original contents while available and use a factual description after routine expiration. Deleting the attachment removes its original and description from future attachment-based use, but does not rewrite earlier conversation messages or responses.
Consolor may maintain user-facing memory within individual Companion relationships, including goals, support preferences, helpful context, and approaches you have said work. Available settings allow you to review, correct, or clear individual Companion memory and to Start fresh. Separately, Shared Companion Context is optional and off by default. When you authorize it, participating Companions may use eligible historical and ongoing context from other participating Companion relationships. Companions that do not participate remain separate.
You can enable Shared Context later, add or remove participating Companions, turn sharing off, or delete Shared Context. Removing a Companion stops its future access and contributions, but previously contributed information remains in the shared context available to other participating Companions. Turning sharing off pauses sharing while retaining stored context. Deleting Shared Context does not delete original conversations or individual Companion memories, and deleted historical context is not silently rebuilt; a new historical import requires authorization. Limited records may remain under applicable legal, safety, or security exceptions.
Shared Context may preserve meaningful personal experiences, relevant names, and information about third parties. Consolor uses automated tools to mask phone numbers, email addresses, street addresses, account numbers, and similar direct identifiers. Masking is best-effort and may not catch every identifier. Safety-episode information and other content excluded by the Shared Context safety rules are not transferred. Significant shared context does not automatically expire after 180 days solely because of age; retention remains subject to applicable privacy and legal requirements.
3. Third-Party AI Providers
Private Consolor content may be processed by third-party AI providers acting as service providers. Chaperone does not authorize those providers to use private Consolor conversations or consumer health data to train their general-purpose models. Production routing uses the governed provider configuration approved for Consolor, with payload minimization and provider/configuration versioning.
4. Optional Internal Model Improvement
Internal model improvement is separate from the core service. For the initial launch in the available U.S. jurisdictions, this use is OFF unless you separately choose to allow it. Declining does not reduce ordinary access to Consolor.
If you opt in, eligible portions of your Consolor conversations may enter a controlled preparation process. Identifiers and unnecessary metadata are removed or transformed, content is minimized to the approved learning purpose, and only approved appropriately de-identified data is released into the internal improvement dataset.
You may turn this permission off later. Once appropriately de-identified information has been finalized into an improvement dataset or used in a completed model version, it may no longer be technically linked to your account and may not be reversible. New eligible use stops after the withdrawal takes effect.
5. Memory and Personalization
Consolor may maintain user-facing memory such as goals, support preferences, helpful context, or approaches you have said work. Available settings allow you to review, correct, or clear user-facing memory and to Start fresh when entering a new Companion experience.
Consolor does not durably store inferred diagnoses or sensitive traits as user-facing personal facts merely because they were discussed. Safety, security, experiment, and operational records are governed separately from user-facing memory.
6. Safety Layer
Consolor uses safety controls outside individual Companion personalities. Consolor uses automated safety classification and output validation to identify potential safety concerns and determine when responses should be modified, restricted, or replaced with appropriate safety resources.
Safety handling does not mean a human, 911, 988, or another service has been alerted. Consolor may present approved crisis or emergency resources but does not represent an external contact unless that contact actually occurs.
7. Human Access and Monitoring
Consolor is not a live human support or crisis-monitoring service. Chaperone does not promise that a person is reading conversations as they happen or will intervene. Limited authorized access may occur for security, troubleshooting, privacy requests, incidents, approved data governance, or other operational purposes described in our privacy notices.
8. Your Controls
- Required sensitive/health-data consent before ordinary substantive support processing.
- Separate optional permission for internal model improvement, off by default at launch — granted, declined and withdrawn in privacy and data use settings.
- Ability to turn optional model-improvement use off later, in the same privacy and data use settings.
- User-facing memory review/correction/clear controls where enabled, in memory settings.
- Separate Shared Companion Context controls to authorize participating Companions, manage participation, pause sharing, and delete shared information without deleting original conversations or individual Companion memories.
- Start fresh for ordinary context when switching or restarting experiences.
- Account deletion and privacy-request paths — see Your Privacy Rights.