Are EHR audit logs sufficient for HIPAA compliance of clinical AI?

No. EHR audit logs capture access-level events: who accessed which record, when. They do not capture: what the clinical AI recommended and why, which clinical criteria the AI applied, what model version produced the recommendation, whether a physician reviewer overrode the AI output, or whether any records were altered after the AI decision was made. For clinical AI in scope, a separate AI decision record satisfying HIPAA §164.312(b) is required in addition to EHR access logs.

How long must clinical AI decision records be retained under HIPAA?

HIPAA §164.316(b)(2) requires retention for 6 years from the date of creation or the date when last in effect, whichever is later. OCR extends this expectation to audit logs for ePHI systems. Clinical AI decision records should be retained for a minimum of 6 years. Seven-year retention is increasingly standard to account for state-specific requirements (California, New York, Texas impose longer periods) and regulatory investigation timelines.

Does a clinical AI vendor need a Business Associate Agreement (BAA)?

Yes, if the vendor processes, stores, or transmits ePHI. A clinical AI audit trail vendor that receives decision records containing patient data is a Business Associate under §164.308(b). Covered entities must obtain a signed BAA before the vendor handles ePHI. Deploying clinical AI without a BAA in place means any vendor data breach becomes the covered entity's HIPAA breach notification obligation under the Breach Notification Rule.

What clinical AI use cases trigger HIPAA §164.312(b) audit requirements?

Any clinical AI agent that accesses ePHI is in scope: prior authorization agents (accessing diagnosis codes, CPT codes, clinical notes), clinical triage scoring (processing chief complaint, vitals, EHR data), diagnostic support AI (reviewing imaging reports, lab panels), care gap identification (scanning patient panels), and any LLM-based agent using patient records as RAG context. An AI agent that processes only fully de-identified data (HIPAA Safe Harbor or Expert Determination standard) is not in scope.

What is on-premise deployment and why does HIPAA require it for some organizations?

On-premise VPC deployment means the AI audit trail SDK and storage run inside your infrastructure — ePHI never leaves your network to reach a third-party cloud. Hospital systems and payers with strict data residency requirements use on-premise deployment to eliminate BAA complexity with the audit trail vendor, satisfy data residency obligations, and maintain direct control over all ePHI-containing records. Tenet AI supports on-premise VPC deployment for Enterprise customers.

How does HIPAA §164.312(b) interact with HIPAA §164.312(c)(1) Integrity Controls?

Section 164.312(c)(1) Integrity Controls requires covered entities to protect ePHI from improper alteration or destruction. For clinical AI decision records, this means records must be tamper-evident — any modification after capture must be detectable. Cryptographic signing (SHA-256 hash + Ed25519 signature) satisfies §164.312(c)(1) for AI decision records. Without tamper-evidence, an audit record can be modified after the fact without detection — which violates both §164.312(b) (record and examine) and §164.312(c)(1) (integrity).

Does HIPAA apply to clinical AI deployed by EU healthcare organizations serving US patients?

HIPAA applies to US covered entities and their business associates, not based on where the AI vendor is located but on whether the entity is a covered entity and whether the data is ePHI. A US hospital or health plan serving US patients is a covered entity subject to HIPAA regardless of where the AI vendor is incorporated. A European healthcare organization serving only EU patients is not a HIPAA covered entity — but may be subject to EU AI Act for clinical AI and GDPR for patient data processing.

HIPAA Audit Controls for Clinical AI Agents: §164.312(b) in Practice

Q: What does HIPAA §164.312(b) require for clinical AI agents?

HIPAA §164.312(b) — the Audit Controls standard — is a Required (not addressable) standard under the Security Rule's Technical Safeguards. It requires covered entities and business associates to implement hardware, software, and procedural mechanisms that record and examine activity in information systems that contain or use ePHI. For clinical AI agents, this means capturing what the agent decided, for which patient, based on what data, with what outcome — not just which records were accessed (which EHR logs already cover).

HIPAA §164.312(b) is the Audit Controls standard under the Security Rule's Technical Safeguards. It requires covered entities and business associates to implement hardware, software, and procedural mechanisms that record and examine activity in information systems that contain or use electronic protected health information. For clinical AI agents — prior authorization, clinical triage, diagnostic support, care gap identification — this means audit records must capture not just which records were accessed, but what the AI recommended and why. EHR event logs do not satisfy this requirement alone.

What HIPAA §164.312(b) Actually Requires

Section 164.312(b) Audit Controls is Required (not addressable) under the HIPAA Security Rule Technical Safeguards. The standard requires hardware, software, and procedural mechanisms that record and examine activity in information systems containing ePHI. Two key words: record AND examine. The records must be reviewable in a meaningful way — for clinical AI, that means structured enough to answer: what did the agent decide, for which patient, using what data, with what outcome. Unlike EHR access logs (who accessed which record), clinical AI audit records must capture the reasoning-level event: what the agent recommended and why.

Which Clinical AI Systems Are In Scope

A clinical AI agent is in scope for §164.312(b) if it: accesses patient records as input or context, outputs recommendations based on patient-identifiable data, routes or prioritizes patient cases, processes ePHI as part of RAG pipelines, or ingests ePHI via API for any downstream AI decision. The four most common in-scope agent types: prior authorization agents (accessing diagnosis codes, CPT codes, clinical notes), clinical triage scoring (processing chief complaint, vitals, EHR data), diagnostic support (reviewing imaging reports, lab panels, clinical notes), and care gap identification (scanning patient panels with full record access).

Why EHR Audit Logs Are Insufficient for Clinical AI

EHR audit logs capture access events: user logged in, record accessed, field modified. This satisfies §164.312(b) for the EHR system. It does not satisfy it for a clinical AI agent. When a prior auth AI agent queries a patient record and produces a denial recommendation, the relevant activity is what the agent recommended and why — not just that records were accessed. EHR logs cannot show: the clinical criteria the AI applied, which policy version governed the decision, the agent's confidence and reasoning, whether a physician reviewer overrode the recommendation, or whether the record was modified after capture. OCR investigations into clinical AI outcomes will request all of these.

6-Year Retention and Business Associate Agreements

Two HIPAA requirements frequently missed: (1) HIPAA §164.316(b)(2) requires documentation retention for 6 years from creation or last effective date. OCR applies this expectation to audit logs for ePHI systems — clinical AI decision records must be retained for 6 years minimum, with 7 years increasingly standard to account for investigation timelines and state-specific requirements. (2) Any vendor who receives, processes, maintains, or transmits ePHI is a Business Associate — requiring a signed BAA under §164.308(b). An AI observability or audit trail vendor receiving clinical AI decision records containing patient data is a Business Associate. Deploy without a BAA and the vendor's breach becomes your notification obligation.

Implementation with TenetClient for Clinical AI

Install pip install tenet-ai-sdk. Initialize TenetClient with your API key (cloud) or VPC endpoint (on-premise, for entities requiring ePHI to never leave their network). Wrap each clinical AI decision with tenet.intent() context manager: call intent.snapshot_context() to capture the complete patient context, procedure codes, diagnosis codes, policy version, and model version for post-hoc reconstruction; call intent.decide() to record the recommendation, clinical reasoning, and confidence; call intent.execute() to close the tamper-evident record and return a record_id. For physician reviewer overrides, call tenet.record_override() with the reviewer's pseudonymized ID, original recommendation, and override decision — satisfying both HIPAA §164.312(b) and EU AI Act Article 14 documentation requirements.