Trust & Responsible AI
Trust has to be earned.
Lingwell asks people to put words around experiences that may be personal, difficult, or hard to explain. We take seriously the responsibility that comes with creating a place for those conversations.
Return to Lingwell overviewRead the questionsYour conversations
Your conversations are not an organizational dashboard.
Organizations can offer Lingwell without gaining access to the private conversations people have inside it. An organizational administrator cannot open and read a participant’s conversations.
If a participant wants to bring something from Lingwell into another conversation, with a clinician, counselor, supervisor, educator, advocate, or someone else, they can intentionally share a summary. That choice belongs to the participant.
Organizations may receive administrative usage information and anonymous, aggregate insights such as broader themes, sentiment, or emerging topics. Those insights are designed to show patterns without exposing individual conversations.
Theme-level reporting does not appear until the required privacy threshold has been reached. Lingwell’s current minimum threshold is 10 active users, and organizations can use a higher threshold when appropriate.
Four principles
What guides how Lingwell is built.
Privacy & Data Stewardship
Sensitive conversation and personal information is protected in transit and at rest. Lingwell’s production infrastructure is hosted on Amazon Web Services, with encrypted production storage and additional application-level encryption for sensitive fields.
Private conversation content is not used for advertising.
Sharing Is Intentional
A participant decides whether a summary leaves their private space.
Sponsoring an organization does not give that organization ownership of, or unrestricted access to, participant conversations.
The person remains in control of what they choose to carry forward.
Responsible Use of AI
AI helps Lingwell respond to the language and experience of the individual rather than forcing everyone through the same script. It is a tool within the conversation, not the authority over it.
Lingwell has a Business Associate Agreement with OpenAI for covered API use, and its OpenAI account is configured so API content is not used to train OpenAI’s models.
Lingwell does not use participant conversations to build an advertising profile or train a public-facing AI model.
Human Agency
Lingwell can ask questions, offer language, notice patterns, and help someone explore an experience. It does not decide what that experience means.
The person using Lingwell remains the authority on what fits, what does not, what they want to explore, what they want to keep, and what they want to share.
Control
People should be able to control their information.
Participants can manage their own Lingwell information, including conversations, summaries, and stored memory. Lingwell also supports complete participant deletion on request.
Some deletion functions remove information immediately, while complete participant deletion and backup expiration follow defined system processes. We maintain those distinctions so that deletion can be handled deliberately rather than making promises the technology cannot keep.
Organizations may also remove participant accounts administratively, but administrative account management does not give an organization access to the participant’s private conversation history.
Safety
Safety without silent surveillance.
Lingwell uses a layered safety system designed to respond more directly as signs of risk become more serious.
The system evaluates multiple categories of potentially high-risk conversation across four levels of escalation. At lower levels, Lingwell may ask clearer or more direct questions. At the highest level, it tells the person to contact 988 or seek emergency help.
Lingwell does not place that call itself. By default, it does not automatically notify an employer, school, healthcare organization, administrator, or emergency service because someone reaches that level.
Certain higher-risk interactions are temporarily logged within Lingwell’s protected systems for safety purposes. Access to those logs is limited to authorized Lingwell administrators, and administrative access to identifiable participant data is itself recorded in an audit log. Safety logs have a defined retention period and are automatically purged; the current default is 30 days unless a record must be preserved for a specific legal reason.
That balance is intentional. A safety system should respond seriously when someone may be in danger without turning every difficult disclosure into an invisible report to the organization that provided the tool.
Clear boundaries
Lingwell is designed for guided reflection.
It can help someone slow down, examine an experience, find clearer language, notice what matters, prepare for another conversation, or understand something they have been carrying.
Lingwell is not a diagnosis engine. It does not provide medical or psychiatric treatment. It does not replace a therapist, physician, counselor, crisis professional, educator, advocate, or other person whose professional judgment matters. And it is not an emergency service.
Healthcare use
How we think about HIPAA and healthcare use.
Healthcare privacy is a responsibility, not a marketing badge.
Lingwell can be used in settings where protected health information may be involved. Those deployments require the appropriate technical configuration, contractual agreements, safeguards, and responsibilities for the particular organization and use case.
Lingwell uses healthcare-appropriate agreements with relevant service providers, including a Business Associate Agreement with OpenAI for covered API use. Where Lingwell acts as a business associate for a customer, the appropriate Business Associate Agreement and deployment requirements are addressed as part of implementation.
We do not believe that work is accurately described by simply putting a “HIPAA compliant” badge on a website.
Frequent questions
Questions people ask first.
Can my organization read my Lingwell conversations?
No. Organizational administrators do not have access to private conversation content. They can see information a participant deliberately shares with them, along with permitted administrative usage information and privacy-protected aggregate insights.
If I share a summary, does the organization receive my whole conversation?
No. Sharing is an intentional action involving the summary the participant chooses to send. It does not unlock the participant’s private conversation history.
Does Lingwell use my conversations to train AI?
Lingwell does not use participant conversation content to train its own public AI model. Its OpenAI account is configured so API content is not used to train OpenAI’s models.
Where is Lingwell data hosted?
Lingwell’s production infrastructure is hosted on Amazon Web Services. Sensitive information is protected through encrypted connections, encrypted production storage, and application-level encryption of sensitive fields.
Can I delete my information?
Participants can manage conversations, summaries, and stored memory, and Lingwell supports complete participant deletion on request. Different deletion actions have different technical effects, so Lingwell maintains defined processes for complete deletion and backup expiration.
What happens if I say something that suggests I may be in immediate danger?
Lingwell’s safety responses become more direct as risk increases. At the highest level, Lingwell tells the person to contact 988 or seek emergency help. It does not itself place a call, and by default it does not report the conversation to the sponsoring organization.
Is Lingwell therapy?
No. Lingwell is a guided language and reflection tool. It can complement professional care, but it is not a substitute for clinical treatment, professional judgment, or emergency support.
If you need more
Good technology should make its responsibilities easier to understand, not harder.
We know organizations may need deeper answers before introducing Lingwell to the people they serve.
Lingwell can provide additional technical, privacy, security, data-handling, retention, and contractual information as part of organizational review and implementation.
If there is something you need to understand about how Lingwell works, ask us.