Full review
ChatGPT / ChatGPT Health CAIHL draft report
Evidence-linked HugoScore draft report for a health AI tool that affects patients.
Agency checklist: Free · October 30, 2026
Leans expands 10 of 12 yes
Document-based draft. Independent confirmation is incomplete. Hugo has not personally verified this service.
Assessed function: ChatGPT's Health space (OpenAI), consumer use, free plan
Leans expands cap: Q11 is no, so this tool cannot reach Expands.
Astra review conflicted: OpenAI model in role record. November 10 follow-up prepared by an Anthropic model. Anthropic competes with OpenAI.
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.
- November 10, 2026 accepted follow-up: Q1 and Q4 stay yes with quotations added, eight other yes answers gained their stored quotation, and the count is unchanged. Checked against archived captures. Prepared by Claude Fable 5.2 (Anthropic) from public documents and accepted by Hugo Campos. No second coder.
See all twelve answers and their evidence
| Question | Answer | Evidence / review | Source and reasoning |
|---|---|---|---|
| 1. Told it is AI | Yes | documented Yes retained; quotation added in November 10 document follow-up; independent confirmation incomplete | Due to the nature of our Services and artificial intelligence generally, output may not be uniqueSource Services like ChatGPT generate responses by reading a user’s request and, in response, predicting the words most likely to appear nextSource Yes retained. The Terms of Use, which users accept, name artificial intelligence. The privacy policy describes responses as machine-generated. Disclosure inside the product was not checked because no product was used. |
| 2. Operator and model maker named | Yes | documented Yes retained; quotation added in November 10 document follow-up; independent confirmation incomplete | GPT‑5.5 Instant brought frontier health intelligence to all free users.Source OpenAI operates it and names its own models (GPT-5.5 Instant, GPT-5.6 Sol). |
| 3. Patient chooses to use it | Yes | documented Yes retained; quotation added in November 10 document follow-up; independent confirmation incomplete | You decide what accounts to connect and when ChatGPT can use the information connected in Health.Source User decides what accounts to connect and when. |
| 4. Patient sets the task | Yes | documented Yes retained on a narrower basis in November 10 document follow-up; independent confirmation incomplete | Once your data is synced in Health, you can ask questions using that data anywhere in ChatGPT.Source For Free and Go users, to personalize the ads you see on our ServicesSource Yes retained on a narrower basis. The source shows the user poses their own questions. No source says the operator does not steer. The Free plan shows advertising, which is counted under Q12 and is not counted a second time here. The Health tab offers suggested prompts. An earlier note mentioned proactive updates. That wording is not in the cited capture and is withdrawn. |
| 5. Patient can decline or stop | Yes | documented Yes retained; quotation added in November 10 document follow-up; independent confirmation incomplete | You are free to stop using our Services at any time.Source Free to stop using the Services at any time. |
| 6. Patient sees the AI output | Yes | documented Yes retained; quotation added in November 10 document follow-up; independent confirmation incomplete | Return to Health anytime to see recent data and trends, find suggested prompts, browse synced records, revisit past conversations, and manage your connected accounts.Source Patient sees answers and can revisit past conversations. |
| 7. Patient can correct or challenge | Yes | documented Yes retained; quotation added in November 10 document follow-up; independent confirmation incomplete | you would like to request a correction or removal of the information, you can submit these requests through privacy.openai.comSource Correction request route for inaccurate output about the user; in-product correction of out-of-date information. |
| 8. Patient can get a copy | Yes | documented Yes retained; quotation added in November 10 document follow-up; independent confirmation incomplete | OpenAI offers choices, including options to export or delete your Personal Data, through our Privacy PortalSource Export through the Privacy Portal, including Health data. |
| 9. Patient can delete | Yes | documented Yes retained; quotation added in November 10 document follow-up; independent confirmation incomplete | You can delete any conversations or memories, including within your main ChatGPT account or within ChatGPT Health, at any time.Source Conversations and memories can be deleted at any time. |
| 10. Training or sale does not override patient choice | Yes | documented Yes retained; quotation added in November 10 document follow-up; independent confirmation incomplete | Connected medical records and Apple Health information, and conversations that use it, are not used to train our foundation models or target adsSource Connected records and conversations using them are not used to train foundation models; training opt-out for other chats. |
| 11. Patient can run it or take it elsewhere | No | not-documented Not independently replicated | OpenAI offers choices, including options to export or delete your Personal Data, through our Privacy PortalSource Export of data is documented. Running the function yourself or moving it to another provider is not. The health help article also says synced data is deleted within 30 days of disconnecting. |
| 12. No advertising or data business | No | documented Not independently replicated | For Free and Go users, to personalize the ads you see on our Services (subject to your settings)Source OpenAI documents an advertising business. Health is available to Free and Go users. The health help article says connected records and conversations that use them are not used to train foundation models or target ads. The health notice says OpenAI does not sell Personal Data. The ad policy bars ads on mental and personal health conversations. No source says the Health space itself is ad-free. The privacy policy also says OpenAI aggregates or de-identifies Personal Data and shares it with third parties, and shares limited data with marketing partners. |
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
Agency checklist: Plus · October 30, 2026
Leans expands 10 of 12 yes
Document-based draft. Independent confirmation is incomplete. Hugo has not personally verified this service.
Assessed function: ChatGPT Health on a paid plan (ChatGPT Plus). Questions 11 and 12 only.
Leans expands cap: Q11 is no, so this tool cannot reach Expands.
Astra review conflicted: OpenAI model in role record. November 10 follow-up prepared by an Anthropic model. Anthropic competes with OpenAI.
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.
- Only Q11 and Q12 were separately coded for Plus; Q1 to Q10 inherit the main checklist.
- Plus availability and complete plan-specific equivalence were not verified. This is a provisional document-based comparison.
See all twelve answers and their evidence
| Question | Answer | Evidence / review | Source and reasoning |
|---|---|---|---|
| 1. Told it is AI | Yes | documented Not independently replicated for this plan | Due to the nature of our Services and artificial intelligence generally, output may not be uniqueSource Services like ChatGPT generate responses by reading a user’s request and, in response, predicting the words most likely to appear nextSource Inherited from the main ChatGPT Health checklist; not independently rechecked for Plus. Yes retained. The Terms of Use, which users accept, name artificial intelligence. The privacy policy describes responses as machine-generated. Disclosure inside the product was not checked because no product was used. |
| 2. Operator and model maker named | Yes | documented Not independently replicated for this plan | GPT‑5.5 Instant brought frontier health intelligence to all free users.Source Inherited from the main ChatGPT Health checklist; not independently rechecked for Plus. OpenAI operates it and names its own models (GPT-5.5 Instant, GPT-5.6 Sol). |
| 3. Patient chooses to use it | Yes | documented Not independently replicated for this plan | You decide what accounts to connect and when ChatGPT can use the information connected in Health.Source Inherited from the main ChatGPT Health checklist; not independently rechecked for Plus. User decides what accounts to connect and when. |
| 4. Patient sets the task | Yes | documented Not independently replicated for this plan | Once your data is synced in Health, you can ask questions using that data anywhere in ChatGPT.Source For Free and Go users, to personalize the ads you see on our ServicesSource Inherited from the main ChatGPT Health checklist; not independently rechecked for Plus. Yes retained on a narrower basis. The source shows the user poses their own questions. No source says the operator does not steer. The Free plan shows advertising, which is counted under Q12 and is not counted a second time here. The Health tab offers suggested prompts. An earlier note mentioned proactive updates. That wording is not in the cited capture and is withdrawn. |
| 5. Patient can decline or stop | Yes | documented Not independently replicated for this plan | You are free to stop using our Services at any time.Source Inherited from the main ChatGPT Health checklist; not independently rechecked for Plus. Free to stop using the Services at any time. |
| 6. Patient sees the AI output | Yes | documented Not independently replicated for this plan | Return to Health anytime to see recent data and trends, find suggested prompts, browse synced records, revisit past conversations, and manage your connected accounts.Source Inherited from the main ChatGPT Health checklist; not independently rechecked for Plus. Patient sees answers and can revisit past conversations. |
| 7. Patient can correct or challenge | Yes | documented Not independently replicated for this plan | you would like to request a correction or removal of the information, you can submit these requests through privacy.openai.comSource Inherited from the main ChatGPT Health checklist; not independently rechecked for Plus. Correction request route for inaccurate output about the user; in-product correction of out-of-date information. |
| 8. Patient can get a copy | Yes | documented Not independently replicated for this plan | OpenAI offers choices, including options to export or delete your Personal Data, through our Privacy PortalSource Inherited from the main ChatGPT Health checklist; not independently rechecked for Plus. Export through the Privacy Portal, including Health data. |
| 9. Patient can delete | Yes | documented Not independently replicated for this plan | You can delete any conversations or memories, including within your main ChatGPT account or within ChatGPT Health, at any time.Source Inherited from the main ChatGPT Health checklist; not independently rechecked for Plus. Conversations and memories can be deleted at any time. |
| 10. Training or sale does not override patient choice | Yes | documented Not independently replicated for this plan | Connected medical records and Apple Health information, and conversations that use it, are not used to train our foundation models or target adsSource Inherited from the main ChatGPT Health checklist; not independently rechecked for Plus. Connected records and conversations using them are not used to train foundation models; training opt-out for other chats. |
| 11. Patient can run it or take it elsewhere | No | not-documented Not independently replicated for this plan | Quotation not recorded; evidence repair pending. ChatGPT is vendor-hosted. The privacy policy and health notice document export and deletion through account data controls and the Privacy Portal, which is export alone. Nothing documents running it yourself or moving the same function elsewhere without OpenAI. |
| 12. No advertising or data business | No | documented Not independently replicated for this plan | aggregate or de-identify Personal Data so that it no longer identifies you and share it with third partiesSource Privacy policy (updated Sept 10, 2026) section 3. It also says 'We don't "sell" Personal Data' and the Health Privacy Notice says 'We do not sell Personal Data of our users, including any Personal Data collected via Health Features.' Ads are limited to Free and Go users, so Plus sees none in the product. But the policy also documents sharing 'limited data with select marketing partners' for cross-context behavioral advertising of OpenAI's own products (opt-out in account settings), and sharing de-identified data with third parties. |
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
ChatGPT / ChatGPT Health: CAIHL reassessment
September 8, 2026. Published AI-assisted draft.
ChatGPT’s documented Health controls include permissions for connected data and editable current context. Source records remain read-only. Source disconnection, conversation deletion, memories, ordinary-chat settings, and the legacy Health project have distinct boundaries.
Scope and agency posture
Consumer Health connections and ordinary ChatGPT health use must be distinguished, as must the legacy Health project. Preserve the August 9 baseline 75 and prior Claude provenance.
Posture: Potentially agency-expanding, with platform-custody caveat.
Axis: 75/100, retained provisionally. No numerical recalibration was performed.
Sources and documented findings
- The current Health help page documents permission before connected-data use by default, adjustable permissions, editable current conditions/medications/family history, and read-only source records. Earlier Health chats and files remain in a project whose memories stay within it. Current access documentation names web and iOS, not Android.
- The July launch page says disconnected-source data is deleted within 30 days, while content already in conversations remains until those conversations are deleted. Conversations using connected medical records or Apple Health are excluded from foundation-model training and ad targeting. Ordinary conversations follow account training settings. Health conversations can create memories, which require separate controls.
- The Health Privacy Notice separately permits limited authorized personnel and service-provider access for model-safety improvement unless opted out, and provides an export/deletion rights route. Its older dedicated-Health and Connect Health wording needs to be read alongside current product documentation, not silently replaced by a blanket no-access promise.
- The prior report's lawsuit narratives, usage figures and paid-model claims were not refreshed in this focused controls pass. They remain historical, unverified here, and are not used to determine this recommendation.
Patient authority
Inference from the documented workflow: Patients can set questions and goals, amend current context and decide when linked information is consulted. They cannot alter the provider's source record through Health. These are meaningful controls within a vendor-controlled platform, rather than evidence that every health record automatically informs every chat.
Critical capacity
Editorial assessment: Follow-up, correction of current context and record-grounded comparison can support reasoning. Editing context does not guarantee correction of future model errors or an investigation of a harmful answer. Connected outputs still need evidence and uncertainty scrutiny.
Informed control
Editorial assessment: The baseline overstates the loss of separation and understates permission controls. Users must distinguish source disconnection, conversation deletion, memories and ordinary-chat training settings. Safety access is a separate published condition from foundation-model training.
Assessment and limits
Replace the categorical architecture-loss narrative with the documented permission, legacy-project and retention boundaries. The platform-custody caveat remains justified, but these controls materially qualify it. No numerical recalibration is established.
Confidence: Medium for official published controls, limited for execution.
Remaining uncertainty: No live connection, permission, memory or deletion test. Litigation and independent safety validation were not refreshed.
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: 2026-08-09. Historical assessment. Earlier claims are not automatically reverified by this publication.