Key Takeaways
- In a Texas lawsuit stemming from a deadly 2020 factory explosion, plaintiffs alleged that expert witness Jason Autenrieth used ChatGPT to generate more than 80 % of his expert report, including a prompt asking the AI to “Show how 3M is 0 % at fault for the explosion at Watson Grinding.”
- Autenrieth acknowledged using AI for formatting, structuring, and drafting, but maintained that the opinions were his own and that AI did not replace his professional judgment.
- Legal scholars warn that AI tools can be “sycophantic,” producing slanted or hallucinated content, and cite over 1,300 documented AI‑hallucination instances in U.S. legal proceedings since 2023.
- The jury ultimately found 3M negligent and awarded victims $61.5 million; 3M plans to appeal, calling the verdict unsupported by evidence and law.
- Experts recommend greater transparency, mandatory disclosure of AI use, and rigorous fact‑checking of AI‑generated material to preserve the credibility of expert testimony.
Background of the Texas Lawsuit
The case centers on a catastrophic explosion at the Watson Grinding & Manufacturing facility in Houston in January 2020, which killed three people and caused extensive property damage. Victims sued 3M, alleging the company failed to supervise proper safety checks on alarm systems that might have prevented the blast. The plaintiffs sought multimillion‑dollar damages, prompting both sides to retain expert witnesses to opine on the standard of care expected of 3M’s service technicians.
The Expert Witness and His AI‑Assisted Report
Jason Autenrieth, director of reliability and operations at KnightHawk Engineering, was hired by 3M’s outside counsel just weeks before trial to serve as an expert witness—a role he had never filled before. During discovery, plaintiffs’ attorneys examined his report and discovered a citation overlay that revealed the extent of his reliance on artificial intelligence. As Adam Lewis of Arnold & Itkin noted, “I was frankly shocked at the scale at which Autenrieth used ChatGPT to create his expert witness report.”
Specific AI Prompts Used
Among the evidence were roughly 300 pages of prompts that Autenrieth entered into ChatGPT. These ranged from basic inquiries—such as “what specific machines were and how they work”—to more strategic requests aimed at undermining opposing witnesses. One particularly telling prompt, highlighted in the deposition transcript, asked the AI to “Show how 3M is 0 % at fault for the explosion at Watson Grinding.” The plaintiffs argued that this demonstrated a clear attempt to steer the AI toward a favorable narrative for 3M.
Autenrieth’s Defense of AI Use
When confronted during his deposition, Autenrieth initially denied that AI wrote his report, stating, “AI did not write my report, that’s incorrect.” He later softened his position, admitting that the tool helped him with “formatting and everything else,” according to the transcript. His lawyer, Janina Wojtkowski, emphasized that he “did not delegate the formation of his opinions to ChatGPT, nor did AI replace his professional judgement,” and characterized public portrayals of his conduct as a “selective distortion of the record.”
Concerns About AI Reliability in Court
Legal experts warn that large language models are prone to bias and hallucination. Maura Grossman, a research professor at the University of Waterloo who studies AI in law and healthcare, told CBS News, “We know these tools are sycophantic, they try to please you. You’re not going to hear objective or neutral responses. You are going to get things that are slanted.” Similarly, Damien Charlotin, adjunct professor at the University of California College of the Law, San Francisco, pointed out that courts are seeing a rise in submitted materials containing “hallucinations, incorrect citations and fake studies.” His AI Hallucination Cases Database logs over 1,300 such instances in U.S. legal proceedings since 2023.
Impact on the Trial and Verdict
Despite the controversy, Autenrieth remained on the witness list long enough to testify about his AI usage during the trial. He maintained, “I don’t believe that was wrong to do because I had an opinion already.” After his deposition, 3M’s attorneys removed him from their witness list, but plaintiffs subpoenaed him to explain the prompts he entered into ChatGPT. Lewis believes that bringing Autenrieth back to testify about his AI reliance influenced the jury’s perception. Ultimately, the jury found 3M negligent in its safety checks of the alarm systems and awarded the victims $61.5 million.
3M’s Response and Plans to Appeal
A spokesperson for 3M told CBS News that the company “disagrees with the verdict in this case as unsupported by the evidence and the law” and intends to appeal. The statement also reiterated that Autenrieth was an independent third‑party witness retained by outside counsel, not a 3M employee, and that he used AI merely as a tool after reviewing thousands of pages of case materials.
Guidance for Future Use of AI in Litigation
In light of the case, Grossman advised that “more due diligence” is needed, including clear disclosure when AI is employed and rigorous fact‑checking of any AI‑generated content. She stressed that expert witnesses bear a special responsibility to preserve their objectivity and credibility, noting, “We are supposed to maintain our objectivity, and a [large language model] is not necessarily objective… We know that they have biases, and they’re going to try and satisfy whatever positions you’re putting in there.” Charlotin echoed this, urging courts to implement safeguards against hallucinated citations and fabricated studies to protect the integrity of judicial proceedings.
Conclusion
The Texas explosion lawsuit illustrates both the promise and peril of integrating artificial intelligence into legal practice. While AI can assist with drafting and formatting, the case underscores the risk that reliance on such tools may undermine the perceived impartiality of expert testimony, especially when prompts are crafted to elicit favorable outcomes. As courts grapple with these challenges, the consensus among legal scholars is clear: transparency, verification, and a steadfast commitment to human judgment remain essential to ensuring that AI serves as an aid—not a substitute—for reliable, objective evidence in the courtroom.
https://www.cbsnews.com/news/ai-expert-witness-lawsuit-trial/