Generative AI Is Reshaping EDiscovery: Risks and Realities

According to Relativity’s Sam Bock, in-house legal teams that haven’t realized generative AI is reshaping eDiscovery need to integrate it into their disposition and preservation practices now. At Relativity Fest Chicago 2024, a panel of experts explored the challenges in managing AI-generated data in the session “Scaling the Cliffs of AI Insanity: Managing Data Created by Generative AI.”

The panel featured speakers from Relativity, Microsoft, Covington & Burling, and Day Pitney. They explored such questions as where the data falls on the spectrum of retention schedules, who owns it, and how it can be accessed and retrieved. Todd Itami of Covington & Burling acknowledged that these are classic information governance framework problems.

Itami indicated information governance is still applicable even though AI introduces new complexities. For instance, regulatory obligations can dictate data retention, yet retention guidelines can also be created to preserve AI interaction histories. Questions about ownership of AI-generated data were also raised, such as whether AI is the custodian of its responses despite human custodians creating the prompts. The panel’s consensus was that humans were the custodians rather than AI.
There are major challenges to how generative AI is reshaping eDiscovery. For example, the persistent nature of AI-generated data complicates data collection and assessment. The data created by AI is retained permanently, allowing AI to access and build on previous information. “A lot of generative AI capabilities are leveraging data that already exists. They’re just giving you an efficient way of surfacing insights. But they’re also creating more content that we have to deal with. Prompts are persistent, responses are persistent,” said Microsoft’s E.J. Bastian. 

Understanding where data resides is another challenge, whether in fragmented transcripts or chat logs. This particularly affects AI-generated transcripts that may lack context or accuracy over time. According to Bastian, “If the data persists, it should be discoverable and respect preservation settings. It should be returned in search and replicate the original experience.”