10 Key Takeaways from eDiscovery Day 2025: Navigating AI and eDiscovery
The 11th annual eDiscovery Day centered on a transformative theme: AI, Risk, and the Future of eDiscovery. Across five expert-led webinars, judges, corporate counsel, and technologists shared a clear consensus—Generative AI (GenAI) is firmly embedded in legal workflows, but its immense power must be balanced by foundational competence, robust governance, and unwavering human oversight.
For legal, IT, and information governance professionals, now is the time to shift from contemplating AI or using it without guardrails to actively governing its use.
Here are the ten most critical takeaways for future-proofing your eDiscovery processes.
1. Prioritize Technical Competence Over AI Literacy
The starkest warning from the ACEDS-sponsored "Walk Before You Run" webinar was that foundational technical competence must precede generalized AI literacy. Without a solid grasp of core eDiscovery and IT systems, GenAI introduces unnecessary risk. As eDiscovery technologist Jennifer Williams, CEDS, Sr. Director, eDiscovery, Litigation Support, Vinson & Elkins LLP advised: "If you can't actually perform a calculation in Excel, if you cannot use some of the basic technology in a truly foundational way, then maybe pump the brakes a little bit on your generative AI usage."
Check out the full webinar here.
2. Human Oversight is the Non-Negotiable Mandate
The "Human-in-the-Loop" requirement remains the ultimate defense against GenAI errors and hallucinations. Experts agreed that AI is merely a tool, and lawyers retain full professional responsibility for any output presented to a court or client. Hon. Andrew Peck underscored this ethical and practical necessity in the "Smarter, Faster, Defensible" webinar: "I cannot imagine an important case and you're not reading the cases you're citing to the court... that's a fault of bad lawyering, not the technology."
3. AI Defensibility Demands New Documentation Standards
To withstand judicial scrutiny, teams must move beyond simply documenting search terms. GenAI adoption requires new, detailed standards for validation and documentation. This involves tracking and documenting the technology version, the authority for its use, and the validation methodology applied.
This documentation, Judge Peck noted, must be done contemporaneously to the work: "Don't think that six months or a year later when what you've done may be challenged, you're going to be able to recreate everything that was done."
4. Contain the Risk of "Shadow AI" Data Leakage
The unapproved use of public GenAI tools is a major emerging risk. Inputting confidential or privileged client data into public LLMs risks data leakage and loss of privilege, potentially training the public models. Legal professionals are advised to integrate AI policies into existing strong information security policies and acceptable use agreements. As Dawn Martin advised in the "Hybrid Work, Hybrid Risks" webinar, "I think just saying, 'Use AI responsibly' is not sufficient in order to make sure that you are mitigating risks involved with all of the things that we've been mentioning so far."
5. Hyperlink Files (Modern Attachments) Demand a Proactive Strategy
The shift to the cloud has made hyperlinked files a massive collection and preservation challenge. You must now preserve the "contemporaneous version" or point-in-time snapshot that existed when the link was created. Greg Buckles noted that in Microsoft 365, channels and groups act as "artificial custodians," requiring a shift from thinking solely about individual employees to identifying and preserving all relevant group data stores to ensure completeness: "What they've done is create artificial custodians that are these group names that you've got to go out and find in order to lock it down there."
Future-proof your eDiscovery process with tips from this webinar brought to you by eDiscovery Today.
6. Messaging and Mobile Data are the New Data Tsunami
Chat and collaboration applications are the fastest-growing source of ESI, creating complex narratives that span multiple platforms . This growth highlights the complexity of BYOD policies and dealing with platforms like WhatsApp, which is widely used overseas. The challenge is often compounded by Microsoft: Teams chat transcripts and Copilot records may be off by default and unsearchable/unpreservable unless explicitly enabled. As Jerry Bui explained, "Even though you have things on hold, they don't put the logs on hold. I think that's a major misunderstanding because the logs will still expire."
7. GenAI Will Accelerate the End of the Billable Hour
GenAI's efficiency in automating repetitive tasks—like drafting, summarization, and first-pass review—is accelerating the shift toward alternative fee arrangements (AFAs). In this environment, legal teams are tasked with demonstrating new value through strategic insight rather than hours spent reviewing documents.
8. Invest in a Cross-Functional Technology Team
Successfully navigating the hybrid risks of a modern data environment requires seamless cooperation among legal, IT, and InfoSec teams. A cohesive team can vet new AI tools and align organizational policies. As Hon. Andrew Peck noted, you want to ensure: "You don't want certain platforms or devices to be onboarded without eDiscovery counsel having a say in the matter."
9. The Door is Open for Authentication Challenges and Deepfakes
The ability of GenAI to create convincing synthetic media (deepfakes) is a critical forensic challenge. The case Mendones v. Cushman & Wakefield Inc., where a deepfake video led to the entire case being thrown out, was a major inflection point. The new burden on forensic investigators is complex, requiring new methods to "really untangle whether or not it's the operating system that's making the changes to this metadata or the user, which has been something that we've had to sort out."
10. ESI Protocols are Your Strategic Governance Document
The ESI Protocol is now your core strategic document for governing new data sources and technologies. Stephanie Clerkin advocated for proactively detailing how hyperlinks will be produced and the required metadata fields for chat data. The protocol must establish ground rules up front, before discovery commences.
Learn about risks GenAI might pose to your eDiscovery process in this webinar from Exterro.
How Exterro Can Help
The need for transparency, validation, and a unified platform approach has never been higher. Exterro’s Data Risk Management Platform is purpose-built to address these exact challenges by integrating Data Privacy, Digital Forensics, and eDiscovery in a single ecosystem.
By automating complex legal holds and providing end-to-end eDiscovery in one platform, Exterro helps legal, IT, and compliance teams implement the best practices discussed at eDiscovery Day to move smarter, faster, and more defensibly.
