A group of patients has filed a proposed class-action lawsuit against two of California’s largest healthcare providers, alleging that the organizations used artificial intelligence to record private medical consultations without obtaining patient consent. The lawsuit, filed Wednesday in federal court in San Francisco, targets Sutter Health and MemorialCare over the deployment of an AI transcription tool known as Abridge.
The plaintiffs claim that during medical visits over the last six months, medical staff utilized Abridge AI to capture and process confidential communications. According to the court filing, patients were not given clear notice that their conversations were being recorded by an AI platform, nor were they informed that this sensitive data would be transmitted outside the clinical setting or processed via third-party systems.
This legal challenge highlights a growing tension between the rapid adoption of “AI scribes”—designed to reduce clinician burnout by automating medical notes—and the stringent privacy protections afforded to patients. In a state like California, where recording laws are among the strictest in the nation, the lack of explicit consent could represent a significant legal liability for health systems.
The complaint alleges that the recordings captured “individually identifiable medical information,” which included patient medical histories, symptoms, diagnoses, medications, and other sensitive health disclosures discussed during what patients believed were private consultations.
The rise of the AI medical scribe
From a technical perspective, tools like Abridge are designed to solve a chronic problem in medicine: the “administrative burden.” For years, physicians have complained about “pajama time”—the hours spent after clinic hours typing notes into Electronic Health Records (EHR). AI scribes use ambient sensing and natural language processing to listen to a doctor-patient conversation and instantly distill it into a structured clinical note.
While the efficiency gains are clear, the data pipeline introduces new risks. These systems typically record audio, transmit it to a cloud-based server for processing, and then use large language models (LLMs) to summarize the text. For patients, the concern is not just who is listening, but where that audio is stored and whether it is used to further train the AI models.
Abridge has seen rapid adoption across the American healthcare landscape. Beyond Sutter and MemorialCare, the software has been deployed by several of the most prestigious health systems in the country, including Mayo Clinic, Duke Health, and Kaiser Permanente.
Privacy laws and the ‘two-party consent’ hurdle
The legal core of this lawsuit rests on the distinction between medical privacy (HIPAA) and recording privacy. While HIPAA governs how health information is shared, California Penal Code Section 632 makes it a crime to record a confidential communication without the consent of all parties involved.
In many states, only one person in a conversation needs to know it is being recorded. However, as a “two-party consent” state, California requires everyone to agree. If a healthcare provider activates an AI recording tool without a clear, signed, or verbal agreement from the patient, they may be in violation of state law, regardless of whether the data is eventually encrypted or anonymized.
The plaintiffs argue that the “clear notice” required for such an intrusion was absent, transforming a routine check-up into an unauthorized recording session. This creates a complex predicament for hospitals: the tool helps the doctor, but the process may violate the patient’s statutory rights.
Key Allegations in the Class-Action Suit
| Issue | Alleged Violation |
|---|---|
| Consent | Lack of clear notice before AI recording began. |
| Data Handling | Transmission of private audio to third-party systems. |
| Scope of Data | Capture of identifiable histories, diagnoses, and medications. |
| Legal Basis | Violation of California state and federal privacy laws. |
Industry-wide implications for healthcare AI
This case arrives at a moment when the healthcare industry is racing to integrate generative AI. The promise is a more present physician who looks at the patient instead of a computer screen. However, the “black box” nature of some AI vendors creates a transparency gap.
If the court finds that Sutter Health and MemorialCare failed to provide adequate disclosure, it could force a nationwide pivot in how AI scribes are implemented. Providers may be required to implement more rigorous “opt-in” workflows—such as digital signatures or recorded verbal consent—before any AI tool is activated in the exam room.
the lawsuit raises questions about the contractual agreements between health systems and AI vendors. If the data is “processed through third-party systems,” as the complaint alleges, the question becomes whether those third parties are acting as “business associates” under HIPAA or if the data is being utilized in ways that exceed the original scope of patient care.
Disclaimer: This article is for informational purposes only and does not constitute legal or medical advice.
The case is currently in its early stages in the San Francisco federal court. The next significant step will be the defendants’ response to the complaint, which will likely address whether patients were notified via signage, intake forms, or verbal cues before the use of Abridge AI.
Do you think AI scribes are a fair trade-off for better doctor attention, or is the privacy risk too high? Share your thoughts in the comments below.
