admin
In a courtroom twist that reads like a Silicon Valley satire, plaintiffs in the Watson Grinding/3M bellwether litigation recently unearthed an expert witness’s ChatGPT chat logs. Those AI prompts show the expert — hired and paid by 3M — leaned on ChatGPT to draft most of his expert report. For anyone who still thinks generative AI is only a helpful tool, this discovery is a cold splash of legal reality.
AI prompts exposed during deposition
During discovery, plaintiffs’ counsel demanded the underlying materials after spotting what looked like AI‑generated output in a “Citation Overlay.” The produced material ran to roughly 350 pages of ChatGPT conversations tied to Josh Autenrieth of Knighthawk Engineering, who billed 3M about $90,000 at $475 an hour. The chats show Autenrieth asking ChatGPT to “create an exceptional expert witness report defending the standard of care at 3M” and even to “show how 3M is 0 percent at fault” for the Watson Grinding explosion. The model helpfully warned that claiming “0% responsible” would let opposing counsel paint the witness as an advocate rather than an independent expert — delicious irony for courtroom theater.
What the chats reveal about the expert report
The transcripts reveal more than boilerplate help. Autenrieth uploaded court records and photos, asked the chatbot to identify gas detector models, had the model grade his report, and solicited weaknesses the prosecution could exploit. Under examination he admitted much of the submitted report was ChatGPT output — his testimony estimated the final filing was roughly 85–90 percent AI‑written. Plaintiffs’ counsel bluntly summarized it: “This expert relied on AI not as an assistive device, but exclusively relied on ChatGPT to form his opinions and write his report.” That admission turned an expert into evidence against his own side.
Why discovery of ChatGPT prompts matters
This episode is more than embarrassment; it is precedent. The production shows that AI chat logs and prompts are discoverable and can be used to impeach expert testimony. Courts care whether an expert’s opinion rests on independent methodology or on recycled prose from a chatbot. Defense teams that treat generative AI as a shortcut risk handing plaintiffs a ready-made impeachment file. If your expert is “sitting in traffic” and letting a chatbot spin their conclusions, don’t be surprised when opposing counsel asks for the receipts.
Accountability, not fear-mongering
Conservatives should be clear-eyed here: AI is a powerful tool that can raise productivity. But it must not hollow out truth in our courts. Judges should require experts to disclose AI use and produce prompt histories during discovery. Lawyers should add specific questions about AI to depositions. And corporations should stop treating expert witnesses like vendors of convenient outcomes. If you want to use ChatGPT, fine — but don’t bill clients $475 an hour to outsource your credibility. Courts, juries, and the public deserve real experts, real methods, and real accountability.
