AI Features Terms

These AI Features Terms supplement the Treat Subscription Services Agreement (the "Agreement") and apply to your use of any feature of the Services that generates content, summaries, transcripts, translations, recommendations, scores, or decisions using artificial intelligence or machine learning (the "AI Features"). Capitalized terms have the meanings given in the Agreement. If these AI Features Terms conflict with the Agreement, these AI Features Terms control as to the AI Features.

1. The AI Features

AI Features may include: transcription and summarization of calls and voicemails; call intelligence and follow-up suggestions; an AI assistant that answers questions about your practice data; AI agents that draft or send messages under rules you configure; translation of messages; drafting of claim narratives, treatment-plan explanations, and patient communications; digitization of forms; and analytics, scoring, and forecasts. Treat may add, change, or retire AI Features at any time and may change the underlying models and providers without notice.

2. Nature of AI Outputs

AI Features produce probabilistic output. Outputs may be inaccurate, incomplete, outdated, biased, or fabricated, and may be wrong in ways that are not obvious. A transcript may misattribute a speaker or miss a word that changes meaning. A summary may omit something material. A translation may lose clinical nuance. A suggested code or narrative may not be supportable. Treat does not review AI outputs before they are shown to you or, where you have configured it, sent on your behalf.

3. Human Review and No Professional Advice

3.1 You review before you rely. You are responsible for reviewing every AI output before using it to make a decision, take an action, communicate with a patient or payer, or enter anything into a clinical, billing, or insurance record. An AI Feature that sends a message automatically does so only because you enabled and configured it, and you are responsible for that message as if you had written it.

3.2 Not advice. AI outputs are informational tools and are not clinical, medical, dental, legal, financial, billing, coding, or insurance advice. No AI Feature diagnoses, treats, or determines coverage or medical necessity. Treat is not a healthcare provider and no provider-patient relationship is created through the AI Features.

3.3 Consequential decisions. You will not use AI outputs as the sole basis for any decision with legal or similarly significant effect on an individual, including decisions about eligibility for care, credit, employment, or insurance coverage, without meaningful human review.

4. Disclosure to Callers and Patients

Several states require that a person be told when they are interacting with an automated system or when a communication was generated by AI, and some require it in healthcare settings specifically. You are solely responsible for determining what disclosures Applicable Law requires of your practice, for making them, and for configuring the AI Features accordingly, including any announcement at the start of an AI-handled call and any labeling of AI-generated messages. Treat may provide disclosure text as a convenience; using it does not make Treat responsible for your compliance.

5. Recording and Transcription

AI Features that transcribe or analyze calls depend on call recording. Your obligations for recording consent under the Communications & Messaging Terms apply in full. Do not enable transcription on any line where you cannot obtain the consents Applicable Law requires.

6. Your Data and PHI in the AI Features

6.1 The BAA governs PHI. Where AI Features process PHI, they do so on your behalf under the BAA. Treat uses PHI in AI Features only to provide, secure, and improve the Services for you, as BAA §3.3 permits, and never sells it.

6.2 Model providers. Treat delivers AI Features through third-party model and transcription providers ("Model Providers"). Treat contracts with Model Providers under terms that prohibit them from training their models on your data and that require them to protect PHI as Treat's Subcontractors under the BAA, and Treat configures those services for zero or minimal data retention where the provider offers it. Treat's current Model Providers are listed at https://www.alltreat.io/legal/subprocessors.

6.3 No foundation-model training on PHI. Treat does not use identifiable PHI to train general-purpose foundation models. Treat may use Usage Data, de-identified in accordance with 45 C.F.R. § 164.514, to evaluate, tune, and improve the AI Features, as the BAA permits.

6.4 Your inputs. You will not enter into an AI Feature any information you are not permitted to disclose to Treat under the Agreement and Applicable Law, including records subject to 42 C.F.R. Part 2 without Treat's prior written agreement.

7. Prohibited Uses

You will not use the AI Features to: generate content that is deceptive, discriminatory, harassing, or unlawful; impersonate a person or conceal that a communication is automated where disclosure is required; make automated decisions about an individual prohibited by Section 3.3; attempt to extract, reverse engineer, or reproduce a model or its prompts; circumvent safety controls; or process data of a type Treat has told you the AI Features are not approved for.

8. Ownership of Outputs

As between you and Treat, you own the AI outputs generated from your Customer Data, subject to Treat's rights in the Services, the models, and Usage Data. Because outputs are generated probabilistically, similar outputs may be produced for other customers, and Treat does not guarantee that any output is unique or protectable.

9. Disclaimer

THE AI FEATURES AND ALL OUTPUTS ARE PROVIDED "AS IS" AND "AS AVAILABLE." TREAT MAKES NO WARRANTY THAT ANY OUTPUT IS ACCURATE, COMPLETE, CURRENT, RELIABLE, FIT FOR ANY PURPOSE, OR COMPLIANT WITH ANY LAW OR PROFESSIONAL STANDARD. TREAT MAY SUSPEND OR WITHDRAW ANY AI FEATURE AT ANY TIME. YOUR USE OF THE AI FEATURES AND RELIANCE ON ANY OUTPUT IS AT YOUR OWN RISK.

10. Indemnification and Liability

The limitations of liability in the Agreement apply to the AI Features. In addition to your obligations under the Agreement, you will defend, indemnify, and hold harmless Treat and its affiliates from all claims, damages, fines, and costs, including reasonable attorneys' fees, arising from your use of or reliance on any AI output; any communication an AI Feature sent at your configuration; any failure to make a disclosure Applicable Law requires; any clinical, billing, or coding decision informed by an AI output; and any input you provided in violation of Section 6.4.