By Bill Hartzer · Published · How this site is researched
What happened
The plaintiff alleged that the head of a splitting maul sold by Harbor Freight detached during use and injured him. His expert, Mark Lehnert, holds no engineering degree but, according to the court, has decades of experience designing and manufacturing power tools and more than a dozen patents. Mr. Lehnert opined on how the maul's head should have been secured to its handle.
After completing his report, Mr. Lehnert entered a query into ChatGPT about the best way to secure a hammer head to a handle. The response was consistent with the opinion he had already reached. At the hearing he said he was "quite amazed" that the "ChatGPT search confirmed what [he] had already opined." Harbor Freight moved to exclude his testimony under Rule 702, citing the ChatGPT query among its arguments.
What the court did
In a Memorandum of Decision and Order of April 23, 2025, Judge Gary R. Brown denied the motion to exclude. On the ChatGPT question, the court wrote: "There is no indication that Lehnert used ChatGPT to generate a report with false authority or that his use of AI would render his testimony less reliable."
The court relied on the sequence of events: the query came after the report was written, the opinion rested on the expert's experience, and the conclusion did not depend on the system's answer. The court quoted the District of Minnesota's order in Kohls v. Ellison for the point that problems arise when experts "abdicate their independent judgment and critical thinking skills" in favor of AI-generated answers, and found that this had not happened here.
Where it stands
The expert was permitted to testify. Later proceedings in the case were not located for this entry.
The obligation it engaged
This matter illustrates Clause 1, sequence, being met. The expert formed and recorded his view in a written report before he consulted the system, and the system's answer was not the basis of the opinion. That ordering is what allowed the court to separate the opinion from the tool.
Clause 2, non-delegation, was satisfied on the same facts, since the conclusion was the expert's own. One point remains for practitioners. A confirming answer from a chat system adds no independent weight under Clause 3, because the system is not an independent source. The court did not treat the query as support for the opinion, and a careful expert would not present it as support either.
Clause 7, contemporaneous record, explains why the sequence could be established at all. The timing of the query came out through the expert's own account at the hearing. An expert who logs each use with a date and the question asked can prove the same sequence from a document instead of from memory.
Sources
- U.S. District Court, E.D.N.Y. (via CourtListener) · April 23, 2025
Ferlito v. Harbor Freight Tools USA, Inc., Memorandum of Decision and Order - Reason (The Volokh Conspiracy) · April 25, 2025
No Problem with Expert's Using ChatGPT to Confirm His Work
What could not be confirmed
The exact punctuation of the Kohls passage as quoted in Ferlito differs between renderings (with or without a comma after "ready-made"); only the unaffected fragment is quoted here. Subsequent history was not located.
About this entry
This entry was checked against the primary document listed in the sources. The entry for Ferlito v. Harbor Freight Tools USA records conduct and what the tribunal did. It does not characterize a holding and it is not legal advice. Last checked September 11, 2026. More matters like this one are collected under the expert witness matters. If something here is out of date, write to [email protected].