Authentication of AI-Generated Evidence: Interplay of FRE 104, 403, 702, and 901; Proposed FRE 707 and 901(c)

This CLE webinar will address authentication and admissibility of evidence created or enhanced by AI. The panel will consider the interplay among Federal Rules of Evidence 104(a) and 104(b), 403, 901(b)(9), and 702, and best practices when someone seeks to proffer or challenge AI-generated evidence, whether acknowledged or unacknowledged.

Litigators are already contending with materials that have been produced, enhanced, or modified by AI systems: sentiment analysis showing attitudes and preferences in class actions or employment litigation, programs that purport to identify habits and predict next actions, enhanced video or audio in premises liability or property insurance cases, timelines created from voluminous records, accident reconstruction, progression tools to show how people or property look as a condition progresses, underground and underwater visualizations, reconstruction of lost or incomplete contracts, electronic messages, or emojis. The list is literally endless.

Whether attempting to lay a foundation or responding to an objection, the party proffering AI evidence will have to demonstrate to the judge that the algorithm that created the evidence is valid and reliable, which FRE 901(b)(9) states can be done by "describing a process or system and showing that it produces an accurate result." When everyone is aware of and acknowledges that AI was involved, the inquiry often looks like any other test of scientific or technical evidence under FRE 702 and varies in complexity with the complexity of the AI system. Because there has been no standardization or broad consensus on the best procedure, the Federal Rules of Evidence Advisory Committee has proposed a new FRE 707 to address admissibility of "machine-generated evidence."

When material is offered that the proffering party does not acknowledge was "machine-generated," which means that another party has alleged that the evidence was AI-generated, nothing is settled. If the opposing party presents sufficient evidence from which a jury could conclude that the material was AI-generated, then most courts and commentators take the position that FRE 104(b) requires the jury to hear and decide admissibility issues. The consensus among commentators is that this is a bad idea. Proposed FRE 901(c) is intended to address this issue and is likely to be a good first step.

Listen as this panel of expert litigators discusses how courts are dealing with these issues now and offers guidance and practical tips for addressing machine-generated evidence.