HugoScore hugoscore.org

Full review

No Barrier CAIHL draft report

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

Agency checklist · October 30, 2026

Constrains 2 of 12 yes

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

Assessed function: No Barrier AI medical interpretation (real-time speech-to-speech AI interpreter with visual term support and bilingual discharge notes) purchased by institutions and used directly with patients.

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 AINodeployer-dependent
Not independently replicated
obtained any and all permissions, authorizations and/or informed consents from the patients Source

The Terms put patient consent on the customer. No No Barrier-provided notice to the patient is documented.

2. Operator and model maker namedNonot-documented
Not independently replicated
our proprietary AI Medical Interpreter platform Source

Operator named: NoBarrier Inc. / No Barrier AI, Inc. (San Francisco). The maker of the underlying speech and translation models is not named in the pages read. 'Proprietary' is the Terms' wording for the platform. The Trust Center was linked but not read.

3. Patient chooses to use itNodeployer-dependent
Not independently replicated
The care team selects the patient’s preferred language Source

Care team starts the session. Institution-purchased.

4. Patient sets the taskNonot-documented
Not independently replicated
Quotation not recorded; evidence repair pending. Source

The care team sets the language and session. The patient does not set the purpose.

5. Patient can decline or stopYesdocumented
Corrected after independent review; accepted by editor
You shall offer the patients the option of using this Platform or another method Source

Section 7.3 requires offering another method. This documents a patient choice, not tested equivalence of interpretation quality.

6. Patient sees the AI outputYesdocumented
Not independently replicated
As complex medical terms arise, the system displays visual aids helping patients understand conditions and procedures more clearly. Source

The patient sees system-generated visual aids and hears the interpretation, per the vendor's product page (marketing). The bilingual discharge summary is also generated automatically. Vendor page, not a binding document.

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

No route to correct or contest an interpretation is documented.

8. Patient can get a copyNodeployer-dependent
Not independently replicated
you retain ownership rights in and to Your Information, and own the Results Source

'You' is the customer. The patient has no documented route. Content is deleted within 7 days by default.

9. Patient can deleteNodeployer-dependent
Not independently replicated
Every encounter's content is automatically deleted after 7 days, unless otherwise agreed with the customer. Source

Automatic deletion, set with the customer. No patient-initiated deletion. The privacy policy says the customer is responsible for data subject rights for customer information, and it lets customers and site visitors request erasure of their own personal information.

10. Training or sale does not override patient choiceNodocumented
Not independently replicated
on a perpetual basis, and provided Your Information is anonymized, de-identified or pseudonymized, for the purpose of generally enhancing the Platform Source

Terms Section 4 licenses anonymized Patient Records to No Barrier perpetually, with no patient opt-out. The Security page says 'We never train on, repurpose or sell your data'. I answered by the Terms and noted the conflict.

11. Patient can run it or take it elsewhereNodocumented
Not independently replicated
The Platform is licensed and not sold to you under this Agreement Source

Vendor-hosted. No patient route to run it or move the function.

12. No advertising or data businessNodocumented
Not independently replicated
you hereby agree to our collection, use, dissemination, sale, transfer, and exploitation of such Summary Data Source

Terms Section 4 covers anonymous aggregated Summary Data, including 'sale'. The Privacy Policy also documents targeted advertising on the website. The Security page says it never sells data, which conflicts.

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

No Barrier: CAIHL reassessment

September 8, 2026. Published AI-assisted draft.

No Barrier provides institution-purchased medical interpretation. Its terms allow deletion or anonymization and continuing derivative-data uses, while its security page makes broader deletion and no-repurposing promises. Patient alternatives and correction remain deployment-dependent.

Scope and agency posture

Institution-purchased medical interpretation used directly with patients. The July 20 baseline is 43, Mixed, institution-led language-access aid.

Posture: Mixed, institution-led language-access aid.

Axis: 43/100, retained provisionally. No numerical recalibration was performed.

Sources and documented findings

  • Current terms require customers to offer another method and verify results. They allow perpetual improvement use of anonymized, de-identified or pseudonymized information and sale of anonymous aggregate summary data. Identifiable material may be deleted OR anonymized within seven days. The BAA takes precedence for PHI, but was not reviewed.
  • The current security page still promises deletion of encounter content after seven days without exceptions, backups or archives and no training, repurposing or sale. The interpretation page allows a different retention agreement and describes visual aids, bilingual discharge summaries and human interpreters for emotional situations. Those statements are not fully reconciled.

Patient authority

Inference from the documented workflow: Language access can let patients express goals and challenge care decisions. The organization selects and configures the platform. A contractual alternative is meaningful, but it does not establish that a patient can trigger it promptly or correct a disputed translation afterward.

Critical capacity

Editorial assessment: Visual explanation and bilingual notes may improve comprehension. Patient-accessible comparison with source speech, qualified review of important summaries and a durable correction route remain unestablished by this pass.

Informed control

Editorial assessment: The concrete distinction between deletion and anonymization matters to informed participation. Broad no-repurposing messaging does not explain surviving derivative-data permissions or which agreement controls a particular encounter.

Assessment and limits

Keep the institution-led mixed judgment but revise retention wording around the current terms, rather than implying all encounter material is erased. The inconsistency is evidence of incomplete disclosure, not proof of improper processing.

Confidence: Medium for published conditions, low for encounter-level implementation.

Remaining uncertainty: Patient notice, refusal experience, interpreter escalation, deployment-specific BAA and derivative-data handling were not observed.

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: 2026-07-20. Historical assessment. Earlier claims are not automatically reverified by this publication.