Navigating Modern Data and Legal Precedent: A Guide for Outside Counsel

Summary: Recent legal precedents on modern data are reshaping eDiscovery practices while generative AI is poised to cause further disruption. We explore rulings on hyperlinked documents and retention policies and dive into the potential implications of AI-generated content in litigation.

‍

Litigation and eDiscovery attorneys have had a front row seat to the rapid data transformation in the business technology landscape. From WhatsApp messages to Teams and Slack data, to the challenge of disappearing Snapchat messages, outside counsel has had to continuously adapt eDiscovery strategies for well over a decade to account for new file types and data handling.

While the law is starting to catch up, AI and cloud-based data continue to create new legal challenges. As outside counsel, you must keep a pulse on the fundamentals of modern data, and closely monitor how courts are handling these issues to ensure they're adeptly advising corporate clients.

In this article, I’ll unpack recent legal precedents for handling modern data types and go over key considerations for generative AI data that you should consider in order to effectively navigate modern data in eDiscovery.

All in the family? Navigating hyperlinked documents in eDiscovery

The concept of "modern attachments" introduces complexity in document relationships. Rather than functioning as traditional attachments where the recipient receives a copy of a file, the modern attachment behaves like a hyperlink, directing the recipient to the file itself. Changes to a file create versions leading to questions around which version legal teams should be reviewing.

To further complicate matters, preserving or collecting hyperlinked documents in the exact condition they were shared depends upon several factors—like which license your client has for their cloud platform and which export workflows are available.

When advising clients on eDiscovery protocols regarding these issues, attorneys will need to consider the ruling in In re Meta Pixel Healthcare Litig. This case established that hyperlinked documents should not routinely be treated as conventional attachments for preserving "family" relationships in production.

“Accordingly, the ESI protocol should make clear that hyperlinked documents are not treated as conventional attachments for purposes of preserving a ‘family’ relationship in production. However, the Court anticipates that for some documents, it will be important to collect (or attempt to collect) hyperlinked documents and associate them with the underlying ESI in which the links appear. In such circumstances, the parties should consider reasonable requests for production of hyperlinked documents on a case-by-case basis. Such requests should not be made as a matter of routine.”

Slack off at your own risk: The perils of last-minute retention policy changes

Emerging legal challenges with generative AI

The rapid adoption of generative AI tools like Microsoft Copilot and ChatGPT precipitates a whole new set of legal challenges, many of which are the subject of current or recent legal cases. Some of this litigation is likely to become significant precedent in the coming years, and may particularly impact the following subject areas:

Attribution

Generative AI solutions are trained using large quantities of existing content. Whether this is original source material, website content, document sets, lines of code, images, video, or other content, generative AI solutions don’t create wholly original responses to user prompts. Responses are built by synthesizing and refining relevant source material into usually coherent results, making them derivative by nature.

Because the responses produced by generative AI are based on existing material, there is concern about attribution and copyright. The responses do not typically attribute their sources, and it is virtually impossible to trace back to the specific content that influenced the AI’s output. Even when prompted to provide attribution, AI systems can be inconsistent—sometimes accurately identifying sources, sometimes misattributing, and often simply not knowing where specific content came from. This attribution is particularly important when AI tools are used to research and write code that involves open-source projects.

Privacy

Some generative AI solutions are trained using closed or internal data sources. While in theory this seems like it would help ensure data privacy, the lines drawn by companies’ privacy policies become blurry when those datasets are possibly being accessed by users lacking permission to see or use certain records. The more advanced AI tools take permissions into consideration, but this is an emerging technology.

Accuracy

Despite prominent warnings and known inaccuracies, legal concerns remain regarding the privacy of AI-generated content. In one prominent example, Australian mayor Brian Hood considered filing a defamation lawsuit against ChatGPT for falsely claiming he had served time in prison for bribery. While Hood eventually decided not to proceed with the lawsuit, it's possible suits of this nature will emerge more frequently.

eDiscovery scope and disclosures

Generative AI data is changing the conversation around eDiscovery scope because of the data it creates and the novelty of the technology. It is currently unclear whether all the data associated with generative AI, including prompts, responses, and logs, should be considered in scope for eDiscovery. If it is, we will find ourselves in a very familiar place—a mountain of data, most of it irrelevant, and no out-of-the-box capability to preserve, collect, review, and produce it.

Conclusion

By familiarizing yourself with relevant legal precedents and areas of debate, you can guide your clients toward best practices in managing, preserving, and collecting modern data. In doing so, you'll become an indispensable partner for clients in the ever-changing legal landscape.

For more guidance and information on how you can help your corporate clients solve their modern data challenges, check out our information governance page.