No Barrier CAIHL draft report

HugoScore CAIHL Draft Report: No Barrier

• Status: AI-assisted draft for human review
• Last reviewed: 2026-07-20
• Review method: Public-source refresh of official product, FAQ, AI-interpreting, human-escalation, security, privacy, terms, and HIPAA materials, the vendor's comparative study, the 2026 California Health Care Foundation landscape report, current HHS Section 1557 language-access guidance, and the PubMed abstract for a 2026 peer-reviewed patient-preference study. No product access, patient or customer interview, customer contract or BAA review, incident-data review, security-audit inspection, or independent No Barrier accuracy, safety, equity, or outcome validation.
• Service: No Barrier
• Vendor: No Barrier AI, Inc.
• URL: https://www.nobarrier.ai/
• Category: Medical interpretation AI

1. Executive Summary

No Barrier is a vendor-hosted language-access platform used by healthcare organizations during clinical encounters. Its current materials describe real-time AI speech-to-speech interpretation across 295+ language access options, visual supports for medical terms, bilingual discharge notes, and a route to human interpreters. Faster language access can materially support a patient's ability to explain symptoms, understand choices, ask questions, participate in consent, and follow instructions.

Under CAIHL, No Barrier is institution-led but directly patient-facing. Care organizations purchase, configure, and govern the service. The patient experiences its output during a clinical encounter and has no documented control over the vendor, deployment settings, or governance. Its terms require the customer to obtain needed consent and offer the patient another method. A July 2026 vendor article says the system can route to a human interpreter when a patient or provider prefers one, or when a conversation becomes emotional. Public evidence does not show how those protections work at the point of care.

The assessment is therefore mixed, institution-led language-access aid, at 43 of 100 toward patient-directed. The main unresolved questions are patient notice, refusal without penalty, human-interpreter availability, correction after a misunderstanding, review of machine-generated discharge notes, contradictory retention language, and independent product-specific validation. HHS and CHCF sources provide governance context. They do not establish that any No Barrier deployment is compliant, safe, or independently validated.

2. CAIHL Question

Who does this AI serve?

No Barrier directly serves healthcare organizations and care teams seeking fast, scalable language access and lower interpreter costs. Patients with limited English proficiency are direct participants and potential beneficiaries, but they do not appear to select, buy, administer, or independently control the system.

CAIHL classification: institutional, patient-facing AI medical interpretation and language-access platform.

3. What The Service Does

No Barrier describes a SaaS platform for healthcare organizations. The care team selects a language and uses AI for real-time interpretation during in-person, phone, or telehealth encounters. Product materials also describe visual aids for medical terms and bilingual discharge notes. The company distinguishes 295+ language access options from AI coverage for selected languages and human-interpreter coverage for others.

The company says its interface can connect to a human interpreter when a patient or provider prefers one, or when a conversation is emotional. A July 2026 article describes a single in-platform control used by the clinician. This is a vendor statement about intended workflow. The public materials reviewed do not establish whether patients are told about the option, can trigger it directly, or can obtain it without delay or a worse care experience.

4. Patient-Impact Pathway

1. A healthcare organization procures and configures No Barrier. 2. A care-team member selects a language-access method during an encounter. 3. AI interprets spoken communication between the patient and care team. Product materials describe visual aids and bilingual discharge notes in some workflows. 4. The interpretation affects history-taking, consent, diagnosis and treatment discussion, medication instructions, discharge planning, privacy, trust, and the patient's ability to ask questions. 5. The patient may need to seek a human interpreter or another method through the organization. Terms say that option must be offered. Current HHS guidance says covered entities must offer a qualified interpreter when interpretation is requested, accurately and at no cost to the patient. The live No Barrier workflow is not publicly documented. 6. Encounter content may include audio, recordings, transcripts, translations, video, and other session data. One security page claims seven-day deletion with no exceptions, backups, or archives. Newer product material allows a longer period by written agreement, while the terms and privacy policy also describe continued use of de-identified information. 7. Product materials say the platform can automatically create bilingual discharge notes. Current HHS guidance requires qualified human review of critical, complex, or accuracy-essential machine-translated text at covered entities. Public evidence does not establish how No Barrier customers review these notes.

5. Evidence Table

| Source | Evidence | CAIHL relevance |
| --- | --- | --- |
| No Barrier homepage, accessed 2026-07-20: https://www.nobarrier.ai/ | Describes an AI-first language-access platform for healthcare organizations, 295+ language access options, 24/7 use, and organizational operating claims. | Establishes institutional buyer, product scope, and patient-impact pathway. Claims are vendor-authored. |
| No Barrier AI Interpreting, accessed 2026-07-20: https://www.nobarrier.ai/ai-interpreting | Describes care-team language selection, visual supports, bilingual discharge notes, AI and human interpretation, and human escalation connected with a customer's existing provider. It says human interpreters remain better for emotional situations. | Supports potential action support, limits of patient control, and need to verify human escalation in practice. |
| No Barrier healthcare-leader Q&A, updated 2026-07-01, accessed 2026-07-20: https://www.nobarrier.ai/post/healthcare-leaders-questions-ai-interpretation | Says AI and human coverage are combined, describes one-control human routing when the patient or provider prefers it, and says seven-day PHI deletion can be extended by written agreement. | Direct vendor support for the human route and evidence that retention can differ by customer agreement. It is not independent verification of the live workflow. |
| No Barrier FAQ, accessed 2026-07-20: https://www.nobarrier.ai/faq | Describes SaaS access, real-time interpretation, medical terminology, seven-day PHI retention, and internal accuracy claims with stated limitations. | Supports service model, data-retention claim, and the distinction between vendor claims and independent evidence. |
| No Barrier terms of service, updated 2025-10-15, accessed 2026-07-20: https://www.nobarrier.ai/legal/terms-of-service | Defines patient records, places permission and consent responsibility on customers, requires an alternative method, requires human verification, says results can be incorrect, and permits ongoing use of anonymized, de-identified, or pseudonymized information for product improvement. | Central evidence for choice, clinical boundaries, data governance, and customer-controlled deployment. |
| No Barrier privacy policy, updated 2024-02-26, accessed 2026-07-20: https://www.nobarrier.ai/legal/privacy-policy | Describes customer responsibility for customer information and data-subject rights, web tracking, correction and erasure requests, de-identified use, and third-party disclosure. | Shows that general privacy rights and customer-held PHI workflows need to be separated. |
| No Barrier security page, accessed 2026-07-20: https://www.nobarrier.ai/security | Says encounter content is deleted after seven days and that data is not trained on, repurposed, or sold. | Vendor security claim. It must be read alongside the legal documents' de-identified-data provisions. |
| No Barrier Spanish interpreter page, accessed 2026-07-20: https://www.nobarrier.ai/spanish-interpreters | Describes seven-day auto-deletion as admin-configurable and says PHI is not used to train models. | Vendor product-page evidence that the retention period may be configurable, contrary to the security page's no-exceptions language. |
| No Barrier HIPAA page, updated 2024-02-26, accessed 2026-07-20: https://www.nobarrier.ai/legal/hipaa | Describes business-associate, AWS, encryption, and security practices, and tells customers to determine whether a BAA is required. | Supports deployment-specific governance finding. No BAA was reviewed. |
| No Barrier comparative-study article, accessed 2026-07-20: https://www.nobarrier.ai/post/nobarrier-ai-outperforms-traditional-medical-interpreters-in-first-scientific-study | Reports a 91-sentence English-Spanish comparison with external interpreter scoring. | Relevant vendor evidence. The full protocol, data, and independent publication were not found. |
| California Health Care Foundation, *AI and Language Access in Health Care*, 2026: https://www.chcf.org/wp-content/uploads/2026/03/AILanguageAccess.pdf | Maps No Barrier as an AI-first language-access startup, reports 150+ deployment sites and stated multi-layer quality assurance based on interviews and company materials, and recommends risk-tiered human involvement, transparent validation, and community participation. | Independent market and governance context. It does not report an independent No Barrier accuracy, safety, or outcome validation. |
| HHS Office for Civil Rights, Section 1557 language-access guidance, currently linked by HHS: https://www.hhs.gov/sites/default/files/ocr-dcl-section-1557-language-access.pdf | Says covered entities must provide accurate, timely language assistance free of charge, offer a qualified interpreter when requested, and require qualified human review for critical, complex, or accuracy-essential machine-translated text. | Regulatory context for patient choice and bilingual discharge-note review. This report does not make a legal-compliance determination about No Barrier or any customer. |
| Montoya Rubiano et al., NEJM Catalyst, 2026, PubMed record and abstract: https://pubmed.ncbi.nlm.nih.gov/42418606/ | A mixed-methods study with 23 Spanish-speaking surgical patients found that preferences for AI and remote human interpretation varied by clinical context, with AI valued for speed and privacy and human interpretation for emotional or culturally nuanced encounters. | External context for a patient-informed hybrid model. It does not evaluate No Barrier. |

6. Mixed HugoScore Profile

| Public question | Answer | Confidence | Rationale |
| --- | --- | --- | --- |
| Who does this AI serve? | Institution-led, patient-facing | High | Organizations and care teams are documented buyers and users. Patients with limited English proficiency are direct participants and potential beneficiaries. |
| Can patients tell AI is involved? | Partial, deployment-dependent | Medium | Buyer materials name AI and the homepage advertises configurable disclosure. Terms place notice and consent duties on customers, but no patient-facing script, sign, notice timing, or comprehension evidence was found. |
| Can patients meaningfully choose? | Partial, deployment-dependent | Medium | Terms require another method and a vendor article describes one-control human routing. HHS guidance says covered entities must offer a qualified interpreter when requested. Live notice, patient initiation, wait time, and the consequences of refusal are not public. |
| Can patients correct or challenge what the AI produces? | Partial during the encounter, unclear afterward | Low-medium | Authorized users must verify results and a patient may seek clarification live. No public transcript review, correction, complaint, or adverse-event process for patients was found. |
| Does it help patients understand or act? | Potentially yes | Medium | Real-time interpretation, visual supports, and bilingual discharge notes could improve understanding and action when interpretation is accurate, the patient has a meaningful human alternative, and critical translated notes receive qualified review. |

Equity Burden

Coverage claims are broad, but public evidence does not establish performance by language, dialect, accent, hearing or speech disability, literacy, cognitive status, age, pediatric use, environment, or culturally complex conversation. The company acknowledges a role for human interpreters in emotional conversations. The same need may arise in other high-stakes contexts. CHCF recommends validation by language and content type, plus participation from limited-English-proficiency communities. No product-specific, independently reviewed subgroup or patient-experience evidence was found.

Data Governance

The security page says encounter content is deleted after seven days with no exceptions, backups, or archives. Newer product material says seven days unless a longer period is agreed in writing, and another product page describes retention as admin-configurable. Terms and privacy materials separately allow ongoing use of anonymized, de-identified, or pseudonymized information for product improvement and broader use of aggregated information. The available materials do not reconcile these statements or establish what is de-identified, how that status is verified, the applicable BAA terms, provider access, patient export, patient deletion, or current recipients. This is a disclosure boundary, not evidence that the company mishandles data.

Clinical Boundaries

The terms state that No Barrier is not medical advice and should not be the standalone basis for clinical decisions. They acknowledge that AI results may be incorrect, require authorized-user verification, and require another method to be offered. These are important safeguards. Current HHS guidance separately requires qualified human review for critical, complex, or accuracy-essential machine-translated text at covered entities. CHCF recommends risk-tiered human involvement. Public materials did not establish deployment rules for emergencies, diagnosis disclosure, pediatrics, adverse events, mandatory human escalation, or review of bilingual discharge notes.

Evaluation Ownership

Public success measures center on connection time, language coverage, throughput, and cost. The vendor's reported English-Spanish comparison is relevant but not independent product validation. CHCF independently maps No Barrier's market presence and stated quality controls based on interviews and company materials. It does not report an independent product validation. The external 2026 patient-preference study supports a context-sensitive human and AI infrastructure, but it did not test No Barrier. No public patient-partnered No Barrier evaluation, product-specific independent accuracy study, subgroup-performance report, or clinical-outcome evaluation was found.

7. Key Unknowns

• What a patient sees or hears before AI interpretation begins.
• Whether a patient is told about the human option and can refuse AI interpretation or select a human interpreter without delay, pressure, or loss of care access.
• How the stated human-interpreter route works during sensitive, high-stakes, pediatric, cognitive, or legally consequential encounters.
• Whether patients can access, review, correct, export, or complain about transcripts, translations, discharge notes, or related encounter data.
• Whether No Barrier's product-specific accuracy and error patterns have been independently validated across languages, dialects, accents, settings, and clinical tasks.
• How the company reconciles a seven-day no-exceptions statement with longer retention by agreement or admin configuration, and how it distinguishes identifiable-content deletion from continuing de-identified or aggregated-data use.
• Whether machine-generated bilingual discharge notes receive qualified human review when the content is critical, complex, or accuracy-essential.
• Deployment-specific BAA, retention, backup, access-control, incident, and audit-log practices.
• Accessibility for people with hearing, speech, cognitive, and digital-access needs.

8. Patient Agency Interpretation

Language access is not a convenience feature. It shapes whether a patient can give a history, understand risk, question a recommendation, make a decision, and follow a care plan. No Barrier can expand agency when it removes a delay or replaces unsafe informal interpretation with accurate, transparent, patient-chosen support.

The same platform can constrain agency if AI becomes the cheap default without meaningful notice, an equally usable human option, or a way to correct harm. The patient-preference literature supports a hybrid approach where the patient and clinical context help determine whether AI or a human interpreter is appropriate. No Barrier's publicly stated escalation path is promising. It is not sufficient evidence that every deployed patient can exercise that choice.

9. Review Provenance

• Criteria source: HugoScore patient agency framework derived from CAIHL, using the same public questions and mixed answer types applied to every tool.
• Reviewer: AI-assisted public-source draft refreshed in OpenAI Codex. No named human reviewer is recorded.
• Model: OpenAI Codex / GPT-5.
• Human review: No comprehensive human review has been completed or claimed.
• Review date: 2026-07-20.
• Limitations: No hands-on product use, patient-facing workflow observation, vendor or customer interview, contract or BAA review, audit or incident-data review, accessibility evaluation, or independent product-specific validation. Vendor claims and the vendor-published comparison are not treated as independent evidence. HHS and CHCF sources provide governance context, not a compliance determination.

10. Publication Recommendation

Publish as an AI-assisted, source-backed draft only. Keep the agency posture as Mixed, institution-led language-access aid, the axis position at 43 of 100 toward patient-directed, and the confidence as Medium draft. Do not call the profile reviewed or verified.

Human review should prioritize live patient notice, refusal without penalty, availability of human interpreters, patient-accessible correction and complaint routes, review of bilingual discharge notes, reconciliation of retention claims, and independent validation across languages and care settings.