HugoScore hugoscore.org

Full review

Carey CAIHL draft report

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

Agency checklist · October 30, 2026

Mixed 6 of 12 yes

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

Assessed function: Family-facing administrative concierge with AI and human assistance. Patient authority distinguished from a caregiver's authority to act.

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
govern your access to and use of our healthcare and caregiving AI assistant Source

Terms name the service an AI assistant, and the homepage FAQ says it uses AI with human support. The disclosure is in the Terms accepted on use, not a separate in-product AI notice.

2. Operator and model maker namedNonot-documented
Not independently replicated
General Care Inc d.b.a. “Carey” located in the United States Source

Operator is named (General Care Inc d.b.a. Carey, California). No AI model maker is named in the Terms, Privacy Policy, homepage or the DiMe article.

3. Patient chooses to use itYesdocumented
Not independently replicated
Carey is designed for people managing care for themselves or for someone they love. Source

The account holder starts use and may be the patient or a family caregiver. The Terms contain no representation that a caregiver holds authority to act for the patient, so the care recipient's own choice is not documented.

4. Patient sets the taskYesdocumented
Not independently replicated
it does only what families ask it to do Source

Homepage: 'You tell Carey what you need'. DiMe article describes confirming the task with the family before calls begin. The account holder, not necessarily the patient, sets the task.

5. Patient can decline or stopYesdocumented
Not independently replicated
Carey is designed for people managing care for themselves or for someone they love. Source

Self-started: Q3 is yes and no care depends on using Carey. Source is the Q3 source. Terms also say 'If you do not agree to these Terms, do not use the Service.'

6. Patient sees the AI outputYesdocumented
Not independently replicated
You see exactly what was done, step by step. Source

Marketing statement on the homepage, plus a weekly summary of tasks handled. Call recordings or transcripts are not documented as available to the user.

7. Patient can correct or challengeNonot-documented
Not independently replicated
Quotation not recorded; evidence repair pending. Source

No documented route to correct or contest what Carey did or recorded. The CCPA list in the Privacy Policy omits a right to correct.

8. Patient can get a copyNonot-documented
Not independently replicated
Right to Know what categories and specific pieces of personal information we collect Source

Only a California right to know is listed. No export, download or copy of data or outputs is documented.

9. Patient can deleteYesdocumented
Not independently replicated
You may also delete your data by texting RESET (all caps) to Carey. This will delete your data immediately. Source

Also deletion by email to [email protected]. Anonymized data may be retained indefinitely.

10. Training or sale does not override patient choiceNodocumented
Not independently replicated
Improve AI training and system performance Source

Section 4 says anonymized and aggregated data is used to improve AI training and is shared with vendors, research institutions or commercial partners. No opt-out is documented. Google API data is carved out from AI training.

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

Vendor-hosted only. No self-hosting or documented way to move the function elsewhere. The service has also shut down.

12. No advertising or data businessNodocumented
Not independently replicated
Share with vendors, research institutions, or commercial partners Source

Anonymized data may be shared with commercial partners, and the Terms (6.1) repeat this. Policy also says it does not sell personal information or send it to advertising platforms. Marketing 'HIPAA Certified & Compliant' conflicts with Privacy Policy: 'not a covered entity or business associate under HIPAA'. Binding documents used.

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

Carey: CAIHL reassessment

September 8, 2026. Published AI-assisted draft.

Carey describes user-requested calls, scheduling, refill coordination, and human assistance. Delegated help can support a patient’s chosen task. Task approval, mid-task revocation, caregiver authority, and current legal terms remain incompletely verified.

Scope and agency posture

Family-facing administrative concierge with AI and human assistance. Patient authority must be distinguished from a caregiver's authority to act.

Posture: Potentially agency-expanding.

Axis: 76/100, retained provisionally. No numerical recalibration was performed.

Sources and documented findings

  • The current homepage describes user-requested calls, scheduling, refill coordination, status updates and human oversight when needed. It continues to display HIPAA Certified & Compliant wording. https://www.hicarey.com/

Patient authority

Inference from the documented workflow: The described workflow begins with the user's task. Completion of calls can expand practical authority, but the pages do not establish approval boundaries, revocation mid-task or safeguards when family and patient wishes differ.

Critical capacity

Editorial assessment: Status visibility supports coordination. It does not by itself establish an inspectable call record or a way to challenge an incorrect action.

Informed control

Editorial assessment: Current marketing confirms AI and human involvement. The historical legal-versus-marketing tension remains unresolved because current terms did not render, not because bad practice was demonstrated.

Assessment and limits

The qualified positive posture remains plausible for delegated administrative action. Keep its low-to-medium confidence and historical evidence dates. Do not treat a certification label as evidence of patient control.

Confidence: Low to moderate.

Remaining uncertainty: Task-level authorization, human access, revocation and current legal terms 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.