CureWise CAIHL draft report

CureWise: CAIHL reassessment

September 8, 2026. Published AI-assisted draft.

CureWise supports cancer-record questions, adjustable explanations, and trial discovery. Record grounding and match reasons may help patients check options. Agreement between models is not independent validation, especially for treatment guidance.

Scope and agency posture

Patient-directed cancer record interpretation, multi-model comparison and trial discovery, under vendor-hosted processing.

Posture: Potentially agency-expanding, mixed.

Axis: 84/100, retained provisionally. No numerical recalibration was performed.

Sources and documented findings

• The current product page describes revocable record connections, adjustable explanations, shareable answers and trial-match reasons. Its Cross-Check example still synthesizes patient-specific medication-restart guidance, and the large active-trial count remains a vendor claim. https://curewise.com/how-it-works

• The privacy policy documents user-directed sharing, formal correction and deletion requests, retention exceptions and no sensitive-health-data third-party advertising. It does not establish a usable correction workflow for each AI-derived clinical fact. https://curewise.com/privacy

Patient authority

Inference from the documented workflow: Patients choose questions and explanation depth and can take answers to their own care team. The vendor's model selection and synthesis still shape the apparent answer.

Critical capacity

Editorial assessment: Match reasons and record grounding can support inquiry. Model consensus is not independent verification, especially in the concrete treatment example.

Informed control

Editorial assessment: Formal rights and revocation are useful documented conditions. External-provider context, retention and training restrictions remain unresolved from the reviewed material.

Assessment and limits

The qualified positive placement remains supportable. Paid access is a barrier, not evidence of authority loss. The high-stakes treatment example calls for visible uncertainty and clinician checking rather than arbitrary score movement.

Confidence: Low to moderate for public design, limited for accuracy and runtime controls.

Remaining uncertainty: Trial-count interpretation, model data flows, derived-fact correction and independent performance remain unverified

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: 2026-07-10. Historical assessment. Earlier claims are not automatically reverified by this publication.