Duane Morris Takeaways: Given the fast-growing use of generative AI tools such as ChatGPT, much has been written about the privacy issues surrounding these platforms. Many employers have serious concerns about whether sensitive company data risks exposure vis-a-vis the use of chatbots and other artificial intelligence tools. This potential harm was recently cited by parties seeking to protect the confidentiality of sensitive documents produced in federal litigation entitled M.A. v. United Behavioral Health, et al., Case No. 2:20-CV-000894 (D. Utah May 23, 2023). The Court’s decision to ultimately protect the confidentiality of the disputed documents is instructive for businesses that may need to seek similar protection during litigation.
In M.A. v. United Behavioral Health, et al., Plaintiffs asserted causes of action against Defendants, a group of providers, for recovery of benefits under the Employee Retirement Income Security Act (“ERISA”) and violations of the Mental Health Parity and Addiction Equity Act (“MHPAEA”). In response to requests for the production of documents, Defendants designated certain documents as confidential pursuant to the stipulated protective order entered in the case prior to production. Id. at 2. Some of the documents consisted of guidelines for subacute skilled care prepared by a third-party and licensed to Defendants. Id. Plaintiffs challenged the confidential designation and Defendants moved for a protective order. Id. at 3.
Plaintiffs argued that the documents were subject to mandatory disclosure based on provisions of the ERISA and the MHPAEA, and that the mandatory disclosure provisions precluded Defendants from designating the documents confidential. Id. In conducting its analysis, the Court determined that Rule 26(c)(1)(G) governed Defendants’ motion, and the Court applied the three-factor test applicable to motions for a protective order, including: (1) the party seeking protection is required to show that the information sought is a trade secret or other confidential research or commercial information; (2) that such disclosure might be harmful; and (3) the harm from disclosure is outweighed by the need for access. Id. at 3-4. If the moving party can satisfy all three prongs, the burden then shifts to the party seeking disclosure that such disclosure is relevant and necessary. Id. at 4.
The Court’s Ruling
Although Plaintiffs asserted that this balancing test was inapplicable because of the mandatory disclosure provisions contained in the ERISA and the MHPAEA, the Court disagreed. It opined that even if the statutes’ disclosure provisions applied, the question before the Court was not about disclosure (because the documents had already been produced), but whether Defendants were allowed to designate the documents confidential. Id. at 5. Defendants submitted multiple affidavits to demonstrate that the disputed documents contained commercially sensitive, proprietary, and confidential information. Id. at 8. Specifically, regarding the subacute care guidelines, to help illustrate the potential harm, the supporting affidavits established that the care guidelines were proprietary and copyrighted, and further asserted that not only would harm result if competitors could pirate or sell the guidelines if disclosed, but also the proprietary guidelines could be used to “develop artificial intelligence and machine learning models, all of which would allow would-be competitors to use [the] guidelines to develop a broader suite of derivative products and services.” Id. at 10 (footnote 41).
The Court agreed with Defendants and entered a protective order for the disputed documents.
It held that the declarations were “sufficient to establish that unrestricted disclosure of the [documents] would result in commercial and economic injury to Defendants and non-parties.” Id. at 10. Having established the potential harm, the Court further found that Plaintiffs failed to carry their burden of showing that the unrestricted disclosure of the documents was necessary. Id. at 12.
Implications of The Decision
For businesses involved in large, complex class action litigation, the discovery process often involves the production of documents that contain commercially sensitive and proprietary information which merits confidential treatment during the course of the litigation. This decision illustrates that when the opposing side challenges such confidential designations, the party seeking protection will need to demonstrate the potential harm that could result from disclosure.
With the availability and ubiquitous nature of generative AI tools, the potential harm is exacerbated given how AI can further exploit such disclosures. Businesses should be mindful that the arguments that prevailed in this matter could help protect confidential information under similar circumstances.