Mental health AI
Wysa
Wysa’s consumer self-help, institutional referral, and clinician-linked channels have different authority and data-sharing arrangements. This profile distinguishes those channels. The former combined score of 55 is historical, rather than a current score for every deployment.
Checklist assessment dated October 30, 2026. Independent confirmation is incomplete. Hugo has not personally verified this service.
Agency checklist · October 30, 2026
Mixed 5 of 12 yes
Document-based draft. Independent confirmation is incomplete. Hugo has not personally verified this service.
Assessed function: Wysa app: consumer AI Conversation Space (chatbot with third-party LLMs), institutional Digital Front Door and Digital Referral Assistant, and clinician-linked services (Coach, Medical Assistant, Co-Pilot). Operator identity (Touchkin eServices Pvt. Ltd. / Wysa Ltd.) and reporting to institutions differ by service.
Initial Anthropic coding has an Anthropic model in the role record. Astra review conflicted: OpenAI model in role record.
A no can mean missing documentation or institution-dependent control; it does not by itself establish harmful behavior. These categories are not clinical safety or product-quality ratings.
- Public documents do not establish actual product behavior.
- Q9 covers the combined consumer and institutional scope. A consumer-only assessment could differ.
See all twelve answers and their evidence
| Question | Answer | Evidence / review | Source and reasoning |
|---|---|---|---|
| 1. Told it is AI | Yes | documented Confirmed at document level | You are interacting with an AI software when using our AI services, and not a person.Source Global Privacy Policy v7.2.2 (revised 2026-10-16). The Terms also describe LLM-enabled AI. |
| 2. Operator and model maker named | Yes | documented Confirmed at document level | OpenAI, Anthropic LLM, Google Gemini LLMSource Wysa's service provider list names these LLM providers for Wysa apps that provision generative AI. The privacy policy names Touchkin eServices Pvt. Ltd. as 'makers of the Wysa app and its services' and the controller or processor role varies by Institution. |
| 3. Patient chooses to use it | No | deployer-dependent Confirmed at document level | We will work with your public institution, healthcare institution, educational institution or other organization to offer our services.Source Conversation Space is chosen by the user in the consumer app, but Digital Front Door, referral and clinician-linked services are Institution-arranged. Answer is for the whole function. |
| 4. Patient sets the task | No | documented Confirmed at document level | This allows you to control how you want the AI to converse with you.Source The user controls tone through the bot persona. The Conversation Space follows a decision-tree with clinician-written content, and recommendations come from Wysa's closed care library. |
| 5. Patient can decline or stop | No | deployer-dependent Confirmed at document level | Before you can use these Institutional services, you will need to agree to our notices and to the notices of your Institution.Source The consumer app can be stopped, but access to institutional services depends on the Institution agreement and a right to decline AI while keeping the same service is not documented. |
| 6. Patient sees the AI output | Yes | documented Confirmed at document level | These reports give you personalized and useful information based on how you use the app.Source The user sees chat responses and weekly LLM-generated insight reports. |
| 7. Patient can correct or challenge | Yes | documented Confirmed at document level | We change our conversation anytime you inform us that the AI is not helping.Source Documented under 'Automated Decisions'. The policy also gives a 'Fix It' right to correct personal data. No formal appeal of an AI response is described. |
| 8. Patient can get a copy | Yes | documented Confirmed at document level | Get a Copy: You can ask for a copy of the personal data we have about you.Source Also 'Send It Elsewhere' for transfer of personal data. |
| 9. Patient can delete | No | deployer-dependent Corrected after independent review; accepted by editor | Co-pilot service will not be deletedSource For the combined consumer and institutional scope, reset excludes Medical Assistant/Co-Pilot data and retention remains institution-controlled. Consumer conversation deletion alone would qualify. |
| 10. Training or sale does not override patient choice | No | documented Confirmed at document level | This preference data may be used to improve the app's AI conversational models in aggregate and anonymized form.Source Documented use of persona preference data for Wysa's own models. The same policy says conversation messages are not used as training data by the LLM provider and that the AI does not learn on its own. 'Object' rights exist but aggregate anonymized data is outside them. |
| 11. Patient can run it or take it elsewhere | No | documented Confirmed at document level | Send It Elsewhere: You can ask us to send your personal data to someone else electronically.Source Export only. Vendor-hosted and proprietary, so no way to run it or move the function without Wysa. |
| 12. No advertising or data business | No | documented Confirmed at document level | We use Meta Ads Manager to make ads about our service, to send these ads, and to see how well they are doing.Source Policy section on campaigns and promotions. It also says 'We do not share or sell your information, messages, or how you use our apps to advertisers or companies that buy data', and shares aggregated analytics with Institutions. Judgment call for the reviewer: the documented Meta advertising and cohort analysis of 'users' is why this is no. |
Initial coding: Claude Sonnet helpers, compiled by Claude Opus 5.5 (Anthropic). Selective review and corrections: GPT-6 Astra (OpenAI). Editor: Hugo Campos. Acceptance of corrections is not product verification.
Checklist rules · Independent review and limitations · Later evidence follow-up
Earlier v1.3 assessment and control finding
Historical assessment under different rules. It does not set the current checklist zone.
Agency finding, before method v1.3
Mixed, channel-dependent
Legacy editorial placement, pre-method: 55
No eligible reviewer. This tool uses AI models from both companies whose models code HugoScore reviews, Anthropic and OpenAI, so neither may review it. Its finding and any number are editorial placements made before the method existed.
- Evidence
- Documentary review. Public sources only, no hands-on testing by HugoScore.
- Review status
- AI draft. No person has settled the findings.
Earlier evidence and assessment
The sections below preserve the earlier review. Its classifications and provenance are historical; the checklist above gives the current draft classification.
Earlier finding · Legacy editorial placement, pre-method: 55
Method v1.3. The control line is evidence for how the agency finding arose, not a quality score. Legacy editorial placement, pre-method: 55. Not comparable with v1.3 positions.
Mixed, channel-dependent
Wysa’s consumer self-help, institutional referral, and clinician-linked channels have different authority and data-sharing arrangements. This profile distinguishes those channels. The former combined score of 55 is historical, rather than a current score for every deployment.
Patient agency
How this tool changes agency
Exercises can support reflection and action. Clinical improvement is not a substitute for user control over routing, and formal data rights do not establish a direct appeal against a clinical decision.
Actual institutional signup notices, alternatives and review routes require deployment evidence
Patient agency assessment
Who sets and changes the goal?
Self-help users can redirect conversation. Institutional pathways constrain the offered resources, but the policy does not establish that the AI itself denies care or that declining it loses access.
What can the patient understand, question, or do?
Exercises can support reflection and action. Clinical improvement is not a substitute for user control over routing, and formal data rights do not establish a direct appeal against a clinical decision.
Can the patient evaluate the conditions of use?
Nickname-based consumer use must not imply anonymity across referral, WhatsApp or clinical deployments. External LLM no-training statements do not cover all internal improvement uses.
Text findings
Conditions of use
Published controls and their limits
Nickname-based consumer use must not imply anonymity across referral, WhatsApp or clinical deployments. External LLM no-training statements do not cover all internal improvement uses.
What remains unknown?
Not tested or not established
Actual institutional signup notices, alternatives and review routes require deployment evidence
Who evaluated this?
AI-assisted public-source draft
Vendor statements describe published conditions, not independently verified behavior. No clinical, security, accessibility, or legal validation is claimed. Earlier evidence remains dated in the report and history.
Sources checked
Source-specific findings and retrieval limitations are recorded in the full report.
Earlier review provenance
Criteria
CAIHL-derived HugoScore framework and September 7 qualitative review priorities. Draft v1.2 numerical anchors remain unadopted.
Reviewer
AI-assisted public-source reassessment prepared in OpenAI Codex.
AI / model
OpenAI Codex / GPT-6
Human review
Hugo Campos authorized publication of these AI-assisted draft reassessments on September 8, 2026. This does not claim comprehensive human verification of every finding.
Review date
2026-09-08
Limitations
Actual institutional signup notices, alternatives and review routes require deployment evidence No live product use, patient-data upload, account creation, code audit, clinical evaluation, or independent implementation validation.
Earlier review method
Focused public-source reassessment using CAIHL: patient authority, critical capacity, and informed control. Existing evidence plus one focused primary-source pass and at most one targeted follow-up. No live product testing. Numerical scores remain provisional editorial placements, not a new calculation.
Document-based draft · Moderate for documented service distinctions (AI-assisted draft)