A shocking legal case has exposed how an expert witness used ChatGPT to write a biased report defending 3M in a lawsuit linked to a deadly 2020 explosion.
Josh Autenrieth of Knighthawk Engineering was paid upwards of $90,000 by 3M to provide independent expert testimony regarding the incident.
Rather than forming independent conclusions, Autenrieth fed documents, photographs, and his CV into ChatGPT and asked it to construct an expert opinion from scratch.
One of his prompts explicitly instructed the chatbot to “create an exceptional expert witness report and show how 3M is 0% at fault for the explosion at Watson Grinding.”
The explosion occurred in Houston in 2020, killing three people and causing damage to approximately 200 homes in the surrounding area.
Victims sued Watson Grinding, the operators of the plant, after investigators determined the blast resulted from a gas leak traced to a “degraded and poorly crimped rubber welding hose.”
Plaintiffs also brought suit against 3M, alleging the company failed to properly service the gas detection system installed at the facility.
Plaintiffs’ counsel Will Moye uncovered what he identified as a “Citation Overlay” document during discovery and immediately recognised it as a hallmark of ChatGPT-generated content.
Moye requested all AI prompts used in the preparation of the report, a request that paused the deposition for three hours and ultimately produced a 350-page log of exchanges with the chatbot.
Autenrieth had even asked ChatGPT to rate the quality of the finished report, and the tool gave it a score of 97 out of 100.
Despite following the biased initial prompt, ChatGPT itself flagged that claiming “0% responsible” was “an easy target” for opposing counsel, a line that was removed from the final report.
At trial, Autenrieth conceded that “90 to 85 percent” of the report had been written by AI, a remarkable admission that undermined the entire basis of his engagement.
3M subsequently attempted to disqualify its own witness, but Moye called Autenrieth himself to testify, placing the full AI-assisted process directly before the jury.
“This expert relied on AI not as an assistive device, but exclusively relied on ChatGPT to form his opinions and write his report,” Moye told 404 Media in a phone interview.
“He acknowledged [at trial] the prompts he put in were biased toward 3M to help 3M win the case, it’s really egregious,” Moye added, noting many prompts were entered the night before the deposition.
Moye went further, telling 404 Media: “They hired him and he used ChatGPT to write these reports, so really, ChatGPT was the expert in the case. There’s just no question about that.”
Autenrieth himself claimed that “AI helped me to draft a straw man to build off of,” though that framing did little to soften the legal and ethical implications of the process.
The case raises uncomfortable questions for the legal profession about how AI use by expert witnesses should be disclosed, regulated, and tested under cross-examination.
Legal analysts warn that as AI becomes embedded in everyday professional tools, keeping it entirely out of expert reports will become increasingly difficult to enforce or even define.
In the near term, lawyers are advised to instruct their expert witnesses to avoid AI-generated content, but the longer-term challenge of regulating its use remains far from resolved.