Speak Up on AI in Clinical Care - HHS RFI Comments Due February 23, 2026

Reminder: HHS RFI on Accelerating AI Adoption in Clinical Care - Comments Due February 23, 2026

HHS has asked for broad public input on how to accelerate safe, effective use of artificial intelligence in clinical care. The comment window closes February 23, 2026. For legal teams in healthcare and tech, this is a clear chance to influence how federal policy will treat AI tools across regulation, payment, and research.

HHS is seeking concrete, experience-based feedback from builders, buyers, evaluators, users, and patients - including those who want to use AI but face barriers. Brief, specific, and well-cited submissions will carry the most weight.

Background and Purpose

This RFI implements HHS's December 2025 AI Strategy, developed in response to OMB Memorandum M-25-21 and recent Executive Orders on AI. It reflects a department-wide effort spanning FDA, NIH, CMS, and ONC. The agency wants practical recommendations rooted in real workflows, not theory.

Your comment can identify friction points, propose targeted fixes, and help set priorities for near-term policy moves. The more specific your examples, citations, and evidence, the more useful your comment will be.

Three Pillars HHS Can Use to Move AI Into Care

1) Regulation

HHS aims for a risk-proportionate, predictable posture that supports innovation and protects patients. The agency wants to know where current rules slow responsible deployment, where guidance is unclear, and where new flexibilities would help without eroding safety or trust.

2) Reimbursement

Payment policy drives adoption. HHS is asking how to modernize fee-for-service and value-based models so high-value AI tools are accessible, affordable, and not locked behind outdated benefit or coding barriers. Input on competition and affordability is expressly invited.

3) Research and Development

HHS supports a vast research ecosystem and wants ideas to translate AI from concept to clinic. Expect interest in public-private partnerships, CRADAs, and mechanisms that couple real-world testing with sustainable market paths.

Ten Targeted Questions to Address (Answer Some or All)

• What barriers most constrain private-sector AI innovation for healthcare and its use in clinical care?
• Which regulatory, payment, or programmatic changes should HHS prioritize? Include specific citations where possible.
• For non-medical devices, what novel issues arise around liability, indemnification, privacy, and security - and what role should HHS play?
• Which AI evaluation methods (pre- and post-deployment) are most promising, and how should HHS support them (contracts, grants, cooperative agreements, prize competitions)?
• How can HHS support private-sector accreditation, certification, industry-led testing, and credentialing?
• Where have AI tools met or exceeded expectations, and where have they fallen short? Which novel tools could move the needle?
• Who inside healthcare organizations most influences AI adoption, and what administrative hurdles dominate decisions?
• Where would stronger interoperability expand markets and speed development? Call out specific data types, standards, and benchmarks.
• What problems do patients and caregivers want AI to address, and what are their top concerns?
• Which AI research areas should HHS prioritize, and what does current literature say about costs, benefits, and clinical impact?

How Legal Teams Can Build a High-Impact Comment

• Start with your role and environment (health system, plan, vendor, startup) and the clinical scenarios you see (e.g., triage, imaging, RPM, documentation support, risk prediction).
• Define the AI use case, risk profile, and affected populations. Flag safety, bias, transparency, and monitoring needs.
• Identify concrete barriers. Cite rule sections, payment policies, contract terms, or operational bottlenecks that block deployment.
• Propose precise actions: rule changes, guidance, pilots, waivers, coding updates, coverage pathways, or funding programs.
• Recommend evaluation methods: pre-market validation, phased rollouts, post-deployment surveillance, independent audits, and sunset/refresh triggers for models.
• Address legal risk: liability allocation, informed consent clarity, documentation standards, incident reporting, and safe harbors.
• Detail privacy/security controls: de-identification methods, minimum necessary, access controls, audit logs, and secure model lifecycle practices.
• Call out interoperability requirements: specific data classes, APIs, and testing regimes needed to make your use case feasible.
• Include evidence: outcomes, utilization, quality, efficiency, equity impacts, and any published studies or internal pilots.

Regulatory Topics and Citations to Consider

• FDA device scope and software: FD&C Act Section 520(o)(1)(E) (clinical decision support exclusion); premarket pathways; change-control plans for learning systems; postmarket controls.
• Quality systems and software lifecycle: expectations for SaMD development, validation, updates, and real-world performance monitoring.
• HIPAA Privacy/Security Rules: 45 CFR Parts 160 and 164; de-identification, minimum necessary, BAAs, use of third-party models, and model training on PHI.
• 42 CFR Part 2 (substance use disorder records) and state privacy laws that complicate data aggregation for AI.
• ONC Health IT Certification: 45 CFR Part 170; API and data export capabilities relevant to AI workflows.
• Information Blocking: 45 CFR Part 171; clarifications for safety/feasibility rationales, security, content-and-manner, and fees as applied to AI data needs.
• CMS coverage and coding: National/Local Coverage Determinations (42 CFR Part 405, Subpart I); New Technology Add-on Payments (42 CFR 412.88); device pass-through (42 CFR 419.66); RPM/RTM billing pathways.
• Value-based arrangements and data sharing: Stark Law exceptions (42 CFR 411.357) and AKS safe harbors (42 CFR 1001.952) for AI enablement and outcomes-based contracts.
• CLIA (42 CFR Part 493) considerations for AI-enabled diagnostics and lab workflows.

Payment Policy Angles Worth Detailing

• Define when an AI tool is a covered service, supply, or component of a service. Suggest coding or modifier approaches.
• Propose coverage with evidence development where appropriate, with clear data-sharing and patient protection terms.
• Recommend pathways for prospective, bundled, or value-based payment that reward measured clinical benefit - not just software licensing.
• Address affordability: competitive access, pricing transparency, and guardrails against vendor lock-in.

Evaluation and Governance You Can Recommend

• Pre-deployment: risk assessment, clinical validation, human factors review, data representativeness, bias testing.
• Post-deployment: sentinel metrics, drift detection, performance thresholds, adverse event reporting, independent audits.
• Transparency: model documentation, intended use, known limitations, update cadence, and user training.
• Accountability: clear role of the clinician, escalation paths, override documentation, and traceability of recommendations.

Interoperability Priorities

• Standards: HL7 FHIR APIs, USCDI data classes, DICOM for imaging, device data streams, genomic and pathology data.
• Testing: conformance, performance, and security testing that vendors must pass to integrate with EHRs and clinical systems.
• Data portability and procurement: practical terms for export, portability, and avoidance of data silos that stall AI deployment.

Patients and Caregivers

• Value: faster access, fewer delays, clearer information, and targeted support for chronic conditions.
• Concerns: consent, explainability, error accountability, bias, privacy, and opt-out options when feasible.
• Civil rights: compliance with Section 1557 and disability laws; monitoring for disparate impact.

Submission Logistics

File comments through the Federal eRulemaking Portal or by mail/hand delivery. All timely comments will be public at regulations.gov. Do not include Social Security numbers, dates of birth, financial account numbers, personal health information, or proprietary business information.

• Submit at: Federal eRulemaking Portal (regulations.gov)
• RFI notice: Federal Register

Deadline for receipt: February 23, 2026.

Action Plan for In-House Counsel and Policy Teams

• Stand up a small working group (legal, compliance, clinical, IT, reimbursement) this week.
• Select 1-3 priority use cases and map the specific barriers and citations.
• Draft concrete recommendations with proposed text or redlines where feasible.
• Attach evidence (pilots, QI data, published studies) and name the outcomes to track post-deployment.
• Complete legal review, secure executive sign-off, and file before the deadline.

Why This Matters

Rules, payment, and research guardrails set in the next few months will steer clinical AI for years. This RFI is a rare, early venue to influence that direction with specifics drawn from your contracts, pilots, and compliance realities. Make it count - and make it practical.