Behavioral health programs are being offered note drafts, risk flags, and scheduling predictions. Whether those tools are lawful to use depends on the privacy rule that covers the record, the consent the patient has signed, and whether the software is a medical device or a certified decision-support function. Questions to ask a vendor before you turn a feature on are in Evaluating AI in Behavioral Health Software: A Checklist.
When an AI vendor may receive protected health information
HIPAA does not forbid sending protected health information to an AI vendor. A covered entity may disclose protected health information to a business associate, and may allow that associate to create, receive, maintain, or transmit the information on its behalf, if it obtains satisfactory assurance that the associate will appropriately safeguard the information [1]. That assurance must be documented through a written contract or other written agreement or arrangement with the business associate, as 45 CFR 164.504(e) describes [1, 2]. A business associate may use or disclose the information only as permitted or required by its business associate contract or other arrangement under 45 CFR 164.504(e), or as required by law [1]. The contract may not authorize the business associate to use or further disclose the information in a manner that would violate 45 CFR part 164, subpart E, if a covered entity did it, except that the contract may permit the business associate to use and disclose the information for its proper management and administration, as 45 CFR 164.504(e)(4) describes, and to provide data aggregation services relating to the covered entity’s health care operations [1].
HHS applies the same rule to cloud services. When a covered entity or business associate engages a cloud service provider to create, receive, maintain, or transmit electronic protected health information on its behalf, the provider is a business associate, even if it stores only encrypted data and does not hold the decryption key [3]. The covered entity or business associate and the cloud service provider must enter into a business associate agreement [3].
42 CFR Part 2 restricts use and disclosure of a record that would identify a patient as having or having had a substance use disorder, either directly, by reference to publicly available information, or through verification of that identification by another person when a federally assisted substance use disorder program obtained the substance use disorder information after March 20, 1972, or a federally assisted alcohol use disorder or substance use disorder program obtained the alcohol use disorder information after May 13, 1974, for the purpose of treating a substance use disorder, making a diagnosis for that treatment, or making a referral for that treatment [8]. It also restricts using that information, whether or not it is recorded, to initiate or substantiate criminal charges against the patient or to conduct a criminal investigation of the patient [8]. Information obtained before the pertinent date is covered when a part 2 program maintains it after that date as part of an ongoing treatment episode that extends past that date [8]. Since the 2024 amendments, a patient may sign one consent for all future uses and disclosures for treatment, payment, and health care operations [4]. A Part 2 program, covered entity, or business associate may then use and disclose those records for treatment, payment, and health care operations as the HIPAA regulations permit, until the patient revokes the consent in writing [4]. That consent is about disclosure of Part 2 records. It is not a general federal requirement that every patient consent to the use of AI in their care.
Decision support is not a blank check
Section 1557 covered entities must not discriminate on the basis of race, color, national origin, sex, age, or disability in their health programs or activities through the use of patient care decision support tools [5]. They have an ongoing duty to make reasonable efforts to identify uses of those tools that employ input variables or factors that measure race, color, national origin, sex, age, or disability [5]. For each tool identified that way, they must make reasonable efforts to mitigate the risk of discrimination resulting from the tool’s use in their health programs or activities [5].
A Health IT Module certified to 45 CFR 170.315(b)(11) must enable a limited set of identified users to select, meaning activate, electronic decision support interventions, in addition to drug-drug and drug-allergy contraindication checking, that are evidence-based or predictive [6]. The evidence-based interventions use data expressed in the standards in § 170.213, including problems, medications, allergies and intolerances, at least one demographic, laboratory, vital signs, unique device identifiers for a patient’s implantable devices, and procedures [6]. The predictive interventions use any data expressed in those standards [6]. When the developer supplies either kind as part of the module, the module must let that limited set of users access complete and up-to-date plain-language descriptions of the source attributes [6]. For each predictive intervention the developer supplies, the intervention must be subject to analysis of potential risks and adverse impacts associated with validity, reliability, robustness, fairness, intelligibility, safety, security, and privacy; to practices that mitigate those risks; and to policies and implemented controls for governance, including how data are acquired, managed, and used [6]. Source attributes for a predictive intervention include whether it uses the race, ethnicity, language, sexual orientation, gender identity, sex, date of birth, social determinants of health, and health status assessment variables in § 170.315(b)(11)(iv)(A)(5) through (13) as input features, and a description of demographic representativeness for those variables [6].
FDA’s January 2026 clinical decision support guidance clarifies the types of clinical decision support software functions that the criteria in section 520(o)(1)(E) of the Federal Food, Drug, and Cosmetic Act exclude from the definition of a device. Certain clinical decision support software functions that do not meet those criteria are device functions [7]. A program should ask the vendor which category a feature falls in, and should not treat a marketing label as that determination.
What this site says Encore OS does
Encore OS drafts treatment plan goals and interventions with AI, and the clinician reviews and attests to the draft before signing. Drafting runs only after the organization publishes a disclosure for the model, and a patient can opt out. This post does not describe any other AI function as available in Encore OS.
The therapeutic relationship is often described as central to psychotherapy. This post does not cite a meta-analysis for that statement, and it does not treat software as a substitute for the clinician who signs the note.
Sources
- 45 CFR 164.502, Uses and disclosures of protected health information, eCFR, HHS. Accessed September 28, 2026.
- 45 CFR 164.504, Organizational requirements for uses and disclosures, eCFR, HHS. Accessed September 28, 2026.
- Guidance on HIPAA and cloud computing, HHS Office for Civil Rights. Accessed September 29, 2026.
- 42 CFR 2.33, Uses and disclosures permitted with written consent, eCFR, HHS. Accessed September 28, 2026.
- 45 CFR 92.210, Nondiscrimination in the use of patient care decision support tools, eCFR, HHS. Accessed September 28, 2026.
- 45 CFR 170.315, Certification criteria, including decision support interventions, eCFR, HHS. Accessed September 28, 2026.
- Clinical Decision Support Software (guidance, January 2026), FDA. Accessed September 28, 2026.
- 42 CFR 2.12, Applicability, eCFR, HHS. Accessed September 28, 2026.