HugoScore hugoscore.org

Full review

Counsel Health CAIHL draft report

Evidence-linked HugoScore draft report for a health AI tool that affects patients.

Agency checklist · October 30, 2026

Leans expands 8 of 12 yes

Document-based draft. Independent confirmation is incomplete. Hugo has not personally verified this service.

Assessed function: Direct-to-consumer Counsel AI conversation with optional physician escalation. Payer-configured and partner-embedded care are excluded.

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.
  • The independent review screened rules and selected evidence; it did not replicate all twelve answers for this tool.
See all twelve answers and their evidence
QuestionAnswerEvidence / reviewSource and reasoning
1. Told it is AIYesdocumented
Not independently replicated
You will be notified if a particular program you are interacting with through the Services is run through AI Technology. Source

Terms last modified Nov 14, 2025. The home page also says 'Chat with Counsel AI' and the informed consent says the AI is a large language model. I did not see the in-app notice.

2. Operator and model maker namedNonot-documented
Not independently replicated
Quotation not recorded; evidence repair pending. Source

Counsel, Inc. and Counsel Medical Group, P.A. are named. The terms say 'large language models created by third parties' and the privacy policy says 'our third-party Large Language Model partner' without naming the maker.

3. Patient chooses to use itYesdocumented
Not independently replicated
Start free. Add a doctor when you need one. Source

Direct consumer sign-up. The signup flow says 'Start with a free conversation with Counsel AI'.

4. Patient sets the taskYesdocumented
Not independently replicated
Ask Counsel anything at no cost. Source

The user sets the question. Counsel is built to steer toward a paid doctor visit when needed.

5. Patient can decline or stopYesdocumented
Not independently replicated
Start free. Add a doctor when you need one. Source

Self-started tool per the Q3 source, no outside care depends on it. Inside Counsel, the doctor joins the AI chat, so declining the AI while still getting Counsel's physician is not documented.

6. Patient sees the AI outputYesdocumented
Not independently replicated
The AI Technology is a large language model that allows you to ask questions and receive real-time answers and feedback. Source

The user sees the chat. The home page says Counsel remembers conversations.

7. Patient can correct or challengeYesdocumented
Not independently replicated
We invite you to provide feedback if you believe Output is discriminatory or inappropriate for your personal circumstances Source

Also the terms say users can connect with a provider if they doubt output, and the HIPAA notice lets users request amendment of health information. This is a feedback and amendment route, not a guaranteed correction.

8. Patient can get a copyYesdocumented
Not independently replicated
You may request that we provide your health records to you in an electronic format. Source

This is the HIPAA notice for health information held by Counsel Medical. I did not confirm that AI chat transcripts are covered.

9. Patient can deleteYesdocumented
Not independently replicated
User Content that you request to be deleted that is not part of a medical or clinical record will be queued for deletion Source

Takes up to 30 days. Medical and clinical records are not deleted on request.

10. Training or sale does not override patient choiceNodocumented
Not independently replicated
data analysis, training artificial intelligence programs, customer and market research Source

Terms section 11 license covers this. The privacy policy says 'We and our LLM partners may use your data for... improving the quality of our AI models'. No opt-out is documented.

11. Patient can run it or take it elsewhereNonot-documented
Not independently replicated
Quotation not recorded; evidence repair pending. Source

Vendor-hosted only.

12. No advertising or data businessNodocumented
Not independently replicated
allow us and our advertising partners to personalize your experience when using the Services and on third-party platforms Source

The policy also says Counsel may use de-identified data 'for any purpose' and 'disclose such data to any third parties'. The HIPAA notice says Counsel will get authorization before marketing or selling health information.

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 report

Counsel Health: CAIHL reassessment

September 8, 2026. Published AI-assisted draft.

Counsel Health offers consumer AI questions and optional physician escalation, alongside enterprise care pathways. Health-information policy precedence and amendment rights qualify the general terms. Conflicting training descriptions remain unresolved.

Scope and agency posture

Direct consumer AI conversation and optional physician escalation, distinguished from payer-configured and partner-embedded care.

Posture: Mixed, patient-useful access with payer-aligned economics.

Axis: 55/100, retained provisionally. No numerical recalibration was performed.

Sources and documented findings

  • The current homepage offers free AI questions, optional record connection and paid physician escalation. It also markets in-network steering to enterprise buyers. Wisconsin is no longer in its listed general-service exclusions. https://www.counselhealth.com/
  • The privacy policy still permits LLM-partner model improvement, while the FAQ says no AI training. Its PHI section expressly gives precedence to the separate health-information policy. https://www.counselhealth.com/privacy-policy
  • The linked health-information policy documents records access, amendment requests and complaints. It permits specified treatment, operations and research disclosures, with written authorization for other uses and defined marketing or sale. https://www.counselhealth.com/hipaa-statement

Patient authority

Inference from the documented workflow: Direct users initiate questions and can add a physician. The enterprise in-network objective is explicit, but does not show that the direct consumer is forced to accept that route.

Critical capacity

Editorial assessment: Physician escalation offers a concrete challenge route, and records amendment is documented. Neither establishes independent source inspection of every AI answer.

Informed control

Editorial assessment: The training-policy tension persists, but PHI precedence narrows what can be inferred from general terms. Access and amendment mechanisms should replace the baseline's overly broad correction unknown.

Assessment and limits

Mixed remains supported by specific channel incentives and conflicting data statements. Correct current geography and give due weight to documented health-record rights.

Confidence: Moderate for public documentation.

Remaining uncertainty: Which AI chats fall under PHI protections, actual un-escalated oversight, and deployment-specific routing remain unresolved

Published documentation is evidence of stated conditions, not proof of actual implementation. Unknowns did not receive automatic negative points. Funding, sponsorship, and public code do not determine agency by themselves.

Review provenance

  • Reviewer/model: OpenAI Codex / GPT-6.
  • 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.
  • 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.
  • Earlier review: 2026-06-12. Historical assessment. Earlier claims are not automatically reverified by this publication.