Robert Keeling Featured in Legaltech News Article “Georgetown AEDI Day Two: Preserving Privilege and Pushing Back.”

In recent coverage of day two of the 2025 Georgetown Advanced eDiscovery Institute (AEDI), Legaltech News highlights AEDI’s focus on how GenAI-powered tools are raising a range of new questions and issues around attorney-client privilege, work product protections, and data volumes.

One of the day two sessions focusing on GenAI, “eDiscovery & AI: Potpourri,” featured Redgrave partner Robert Keeling providing his perspective and predictions on a mix of AI issues impacting eDiscovery, including: the impact of AI on privilege, whether AI-generated content is a business record and who is the custodian, and how the shift to Agentic AI impacts discovery and liability issues.

Legaltech News writes:

A number of panelists stressed the importance of knowing and understanding the products a company uses and the proactive management of permissions and settings.  Even seemingly benign uses can pose discovery risks.  For instance, most AI summarization tools create transcripts of meetings off of which they base summaries, and depending on their settings may save the transcripts in addition to the summaries, creating a new source of risk.  While these transcripts may not always be admissible as evidence themselves, they can be used to impeach witness testimony in a deposition.

Proactive management of permissions and settings takes on added importance given the frequent creation of new tools and use cases.  Redgrave partner Robert Keeling noted that new products that convert meeting recordings or transcripts into podcast form for users can combine multiple meetings into a single recap, which risks combining privileged and nonprivileged material.

Part of the challenge for in-house teams and discovery attorneys arises from the rapid pace of change in the technologies in question and the eagerness with which businesses are embracing them.  At times, the technology and its deployment move faster than legal departments trying to analyze the risks enterprises and users may be exposed to.

Beyond risks related to privilege and work product protections, AI tools also can create exceptionally large volumes of data that can be challenging to manage.  Keeling noted that, in addition to prompts and outputs, gen AI tools create “artifacts,” byproducts through which they essentially show their work.  A single query in a tool’s deep research mode can result in more than a dozen artifacts, a profusion repeated with every request.  Similarly, podcasts created to recap meetings for users can be made up of hundreds of individual audio files, and can be split back into those individual files on export.

On the whole, there is a growing risk that e-discovery architecture and technology is falling behind the proliferation of new tools and data sources rising in response to business needs and expectations.

Learn more about AEDI and the sessions featuring Redgrave team members here.