Expert witnesses have long stood as the bedrock of evidentiary credibility. They embody “expertise,” lending weight to complex litigation where facts alone are insufficient. The introduction of Artificial Intelligence threatens to erode the judicial
trust in experts that
has been made standard over decades.
[1] Contemporary cases serve as a warning that AI-assisted analysis can backfire.
[2] From fabricated citations to unverified calculations, defense attorneys now face new challenges of ethics, admissibility, confidentiality, and defining “expert judgment” in the modern courtroom.
[3]
Case Study: Kohls V. Ellison
By looking at a particular case, Kohls v. Ellison, one can get a grasp of the precedent for AI interference in expert witness cases. The case from January this year was challenging Minnesota’s ban on AI-generated election deepfakes. During the case a Stanford professor, Dr. Hancock, was hired as an Expert witness. His testimony was subsequently thrown out after it was found that ChatGPT-40 AI hallucinated the citations of his court filing.
The incident not only “shattered credibility” in the immediate case but also signaled a broader judicial skepticism toward AI-assisted expert evidence, inviting greater scrutiny and exclusion from admissible testimony in some courts.
[4] Kohls v. Ellison
underscores the legal community’s deepening unease with generative AI. Fabricated citations can appear so convincingly authentic that even a seasoned academic, an expert in artificial intelligence itself, can be misled into relying on them in sworn testimony condemning AI. In one of the judiciary’s most ironic episodes, Hancock exposes a profoundly troubling erosion of professional competence and evidentiary credibility in the age of algorithmic authorship. Experts are at risk of broader scrutiny due to fabricated citations and the collapse of trust in testimony.
[5] Even highly credentialed experts armed with innovative tools are not immune to the pitfalls of AI hallucination. When the credibility of an expert fails, so too can the case they support.
Broader Implications for Expert Witnesses
Originally a matter of free speech vs. technological regulation, Kohls V. Ellison now serves as a warning about the fragility of judicial trust, and courts are beginning to adapt. In Doe vs. State of Minnesota, plaintiffs sought a full Daubert analysis of both expert declarations. While the court denied a Daubert analysis in full, they did evaluate each expert's competence and credibility by employing Daubert principles
[6]. They also denied a challenge to one expert's declaration on the grounds of sufficient competence and credibility, demonstrating how trust rather than just technical qualification is now critical to determining admissibility.
With higher emphasis on trust comes higher standards for trustworthiness. Hancock’s reliance on unverified AI output led to exclusion, which means courts are aware and on the lookout for AI contamination in expert materials
[7]. The duty to verify and ensure accuracy is both “personal and
non delegable” under the Federal Rule of Civil Procedure 11.
[8] With an evolving jurisprudence, a quiet but significant reinterpretation has been set into motion: the affirmation that testimony must be grounded in verifiable human judgment. No algorithm can absolve a witness or their counsel of that responsibility.
Response
Legal rulemaking and regulatory bodies are taking action. The Maryland State Bar Association recently cautioned litigators that the “use of AI does not relieve personal accountability” from procedural or ethical rules. They go on to say that attorneys and experts must employ “rigorous verification” to AI assisted content.
[9] Legal commentators and academics have developed alike motifs, arguing that reliance on generative AI models in expert testimonies contributes to “emerging risks in legal, academic, and regulatory contexts,” including unvetted output, fabricated citations, and the collapse of institutional trust.”
[10]
Disclosure of AI involvement may soon become a standard expectation in expert declarations and court filings. This shift mirrors trends in early academic and regulatory ethics. For members of the Washington Defense Trial Lawyers (WDTL) and the broader defense community, these developments carry weight. Expert witnesses are integral to defending tort, injury, and product liability cases. Their reports anchor causation, credibility, and damages analyses. The intrusion of AI into this space creates new battlegrounds for admissibility, reliability, and privilege. A single AI-assisted phrase or phantom citation can upend months of case preparation and threaten to erode the credibility upon which expert testimony depends. Defense counsel must therefore vet their witnesses for subject matter expertise and technological discretion. As the boundary between humans and machines continues to blur, the lawyer’s role is to preserve the integrity of judgment itself.
The courtroom’s relationship with artificial intelligence is still being solidified, but one principle remains central: expertise must be earned, not generated.
Lauren Riley is a client relations specialist at MACHAON Medical Evaluations, a core sponsor of WDTL. She is currently pursuing her undergraduate degree on the pre-law track and enjoys gaining real-time insight into defense cases through her work with the medical expert witness firm.
Bibliography
