New Lawsuits Allege AI Violations of Health Care Consent, Notice and Privacy Rules and the Language Access Consequences

In April, two more California health systems were sued in federal court and accused of violating patient privacy, informed consent, and disclosure laws by allegedly using an AI scribe tool to record patient-clinician conversations during medical visits without consent. Sutter Washington v. Health & Memorial Health Services, 4:26-cv-03012-HSG, (N.D. CA, April 8, 2026).

The health system defendants used Abridge, an AI-tool and ambient documentation platform that generates clinical notes from audio recordings. Abridge is not named as a defendant in the lawsuit. According to its website, Abridge adheres to industry-standard privacy and security practices and also recommends that clinicians follow their organization’s disclosure and consent policies. Indeed, Abridge advises health care providers about

“Obtaining patient consent and discussing Abridge with your patient is an important part in the process. Make sure you follow your organization’s recommended guidelines for patient consent.

Here is a sample talk track you can use, when introducing Abridge:

“I will be using a tool that records our conversation to help me write my clinical note, so I can pay more attention to our conversation and less time on the computer. Is that okay with you?”

Ambient AI scribing tools use a microphone-enabled device, such as the provider’s smartphone or tablet, to capture visit interactions and generate editable clinical notes that may improve physician productivity and earnings by freeing physicians from writing their own clinical notes and instead can see more patients.

Patients are seeking class action status and allege that clinicians at Sutter Health and MemorialCare “intercepted, recorded, and processed” audio during their visits without informed consent. Their case claims violations of California state law.

According to the federal complaint:

“Defendants also failed to ensure that patients received clear and conspicuous notice prior to their medical visits that their conversations might be recorded.

Defendants did not require clinicians to follow standardized procedures or scripts to obtain express consent before activating the recording system, nor did Defendants provide reliable visual or auditory indicators notifying patients that recording was occurring during the clinical encounter.”

The Sutter case is strikingly similar to a December 2025 case filed in San Diego County Superior Court, Saucedo v. Sharp HealthCare, (San Diego Sup. Ct., December 2025). According to Health Executive, the case also involves Abridge and alleges health care privacy and consent violation allegations. Chillingly, plaintiffs claim that the defendants noted that patients consented to Abridge use when they did not:

“Sharp HealthCare used an artificial intelligence dictation tool to record patient conversations in secret, in violation of state and federal privacy laws.

According to the complaint, use of the tool—called Abridge—would not be an issue if Sharp sought consent from patients. However, the clinic is accused of marking down that patients consented instead of actually asking.

Staff at the facility have been using the technology since April, the complaint alleges. In covering the story, local KPBS said Sharp HealthCare was using the system to record conversations between patients and clinicians, from which the AI would automatically generate clinical notes for the encounter.

Plaintiffs allege that, in order to perform the note-taking task, the tool must capture everything said in an exam room, including sensitive details on diagnosis and treatment plans—all of which is protected information under the Health Insurance Portability and Accountability Act (HIPAA).

These audio recordings are also allegedly stored by Abridge after use to improve its AI. “

According to KPBS, when Jose Saucedo, the primary plaintiff in the San Diego case, noticed his medical record appeared to be written by AI, he contacted Sharp who confirmed the use of the tool and apologized for the issue.

Perhaps most concerning, Saucedo said the notes contain confirmations that patients were advised about the recordings and consented—affirmations that appear to have been added by the AI itself. KPBS said Sharp declined to comment on pending litigation. Attorneys representing the plaintiff said they estimate 100,000 patient encounters have been recorded since the rollout of Abridge.

Both cases raise significant and troubling issues. They also confirm that many health care organizations embrace AI and implement its use without the proper understanding, training, and policies to use AI compliantly.

According to Medscape Medical News:

“Deirdre Mulligan, JD, professor at the University of California, Berkeley School of Information, who researches responsible AI and technology governance, told Medscape Medical News that health systems are still figuring out how to integrate these tools into clinical workflows while meeting existing consent and disclosure requirements.

If appropriately evaluated, tested, and monitored, and if privacy issues are addressed, these tools can provide more space for interactions between patients and clinicians,” she said. “But to get that process right, organizations must really understand the data flows and have contractual provisions in place to handle patient data and disclosures, especially when the technology is offered by third parties…

Audio recordings central to both lawsuits raise specific [privacy, consent and notice] concerns because they collect patients’ biometric data. “A full audio recording is very different from a physician’s notes,” she [Mulligan] said. “Some patients may want to turn it off, and they should have the ability to say no.

As more and more health systems and insurers adopt AI-driven tools to improve efficiency and reduce administrative burden, legislators and courts must determine how those technologies align with patients’ rights and existing privacy laws.”

Abridge, known as an AI-scribing device, has the ability to capture a greater number and severity of diagnoses that may be particularly advantageous in value-based payment arrangements, translating into higher reimbursements for participating health organizations. In a time of tight health care budgets, layoffs, and hospital closures, higher reimbursements mean increased revenue when such funds are not as readily available as they were just 2-3 years ago.

However, examining the seeming rush to embrace AI-scribing technology, researchers acknowledged that additional spending and revenue driven by AI-generated documentation may not necessarily improve patient care, instead potentially leading to “simply justifying a more lucrative service code for an otherwise-identical office visit.”

Language Access Ramifications:

Although these and other lawsuits do not involve LEP patients and patients with communication disabilities, it is easy to imagine future cases alleging health care civil rights requirements concerning health care providing spoken language, sign language, and Braille accommodations to provide equal access to health care regardless of disability or national origin.

Similar to telehealth’s initial rollout, especially during COVID, hospitals may not routinely provide qualified spoken and sign language interpreters, proficient translations, and Braille and similar documents. In addition, Abridge and similar scribing technology may not factor in language and disability when creating their AI products and their instructional online tutorials for health care providers.

Despite the rapidly growing use of technology such as Abridge, the language and disability aspects of such technology appear not to be getting as much broad attention as is warranted. As lawyers learn in law school, to paraphrase, AI scribing technology and nonexistent, related language access are lawsuits waiting to happen, and they will happen unless the technology and health care systems are aligned to preclude disability, spoken language, and national origin discrimination under federal law.