§ Claim under review · Safety
"In a lawsuit over liability for a Houston explosion that killed three people, Josh Autenrieth of Knighthawk Engineering, an expert witness hired by 3M's legal team, used ChatGPT to create his entire expert witness report defending 3M, asking the AI to 'create an exceptional expert witness report defending the standard of care at 3M' and to 'show how 3M is 0 percent at fault'; a judge ordered 3M's legal team to turn over 350 pages of his ChatGPT logs, which contained public links to his full conversations."
Verdict
Mostly accurate
Confidence
MediumSummary
This is mostly accurate. Reporting by 404 Media, based on court transcripts and discovery records from the Watson Grinding explosion litigation in Houston, found that an engineer hired as an expert witness on 3M's side used ChatGPT extensively to produce his expert report, including prompts asking the AI to write a report defending 3M's standard of care and to show 3M was zero percent at fault. Roughly 350 pages of his ChatGPT conversations were produced to the opposing lawyer, and they included publicly accessible links to his chats. Two details in the viral version are off. No source shows a judge ordering the handover; opposing counsel demanded the prompts during a deposition, which paused for three hours while 3M's lawyers gathered them. And 404 Media described "significant portions" of the report as AI-written, while the witness himself reportedly testified it was 85 to 90 percent. 3M, the engineer, and his firm did not respond to requests for comment, and the underlying court documents were not independently reviewed for this check.
The readings
key figures from the evidenceChatGPT conversation logs produced in discovery
jury verdict for residents, 3M found 30% responsible
Why this verdict
Evidence
The originating report is by 404 Media. Its own lede states that an expert witness in a lawsuit about liability for a Houston explosion that killed three people and destroyed roughly 200 homes "used ChatGPT to write significant portions of his 'expert report,'" that the man was hired by 3M, that he exposed his AI prompts publicly, and that the prompts showed he asked ChatGPT to help him "create an exceptional expert witness report defending the standard of care at 3M," and that the report should "show how 3M is 0% at fault for the explosion at Watson Grinding."
404 Media says court transcripts, deposition documents and discovery records shared with it show extensive AI use in the case.
The mechanism of disclosure is described consistently across outlets and is not a judicial order. Above the Law reports that plaintiffs' counsel Will Moye found the "Citation Overlay" document, recognized it as ChatGPT output, and asked to see any and all AI prompts used in generating the report; the request paused the deposition for three hours and resulted in a 350-page log. Blaze Media, quoting the transcript, reports that the attorney simply searched the document's content and found that Autenrieth's ChatGPT prompts were all available online, asking him "Well, do you realize, sir, that your chat right now is still publicly available?", to which Autenrieth answered "I have no idea how it works," after which the deposition was paused for three hours to gather all the prompts.
On the extent of AI authorship, the trial record is more specific than "entire." Above the Law reports that 3M tried to disqualify its own witness at trial, so Moye called Autenrieth himself, and that in the transcript Autenrieth concedes "90 to 85 percent" of the report was written by AI. Opposing counsel's characterization is stronger: "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.
The report draft concluded that "3M met the applicable standard of care" and was "0% responsible for the January 24, 2020, explosion."
Surrounding facts check out. A Harris County jury returned a $61,578,273.19 verdict on August 10 2026 for a group of residents affected by the January 24 2020 Watson Grinding explosion in West Houston, finding 3M 30 percent responsible, the third jury verdict for community members in four trials involving 3M.
The explosion killed three people, injured dozens and damaged or destroyed hundreds of homes; 3M told ABC13 it disagrees with the verdict, believes it unsupported by the evidence and the law, and will appeal.
Breitbart, summarizing the CSB finding, reports the blast was caused by a "degraded and poorly crimped rubber welding hose" that leaked flammable gas, and that 3M paid Knighthawk roughly $90,000 with Autenrieth billing at $475 per hour.
On subject response: Knighthawk Engineering, Autenrieth and 3M did not respond to 404 Media's requests for comment. No denial and no confirmation of the ChatGPT specifics from any of the three.
Findings
✓ What's accurate 6
- The lawsuit, the explosion, and the death toll are correctly described. The January 24 2020 Watson Grinding blast in Houston killed three people, and 3M was a defendant found 30 percent responsible in an August 2026 Harris County verdict.
- Josh Autenrieth of Knighthawk Engineering was retained as an expert witness on 3M's side, and Knighthawk was paid roughly $90,000, with Autenrieth billing at $475 per hour.
- Both quoted prompts are reported accurately. He asked ChatGPT to "create an exceptional expert witness report defending the standard of care at 3M" and to show that 3M was "0% at fault for the explosion at Watson Grinding."
- Approximately 350 pages of ChatGPT conversations were produced to opposing counsel, and they included publicly accessible links to his full chats.
- He fed the model large volumes of case material. Outlets report he uploaded hundreds of court records and asked the AI to read them and produce a defense of 3M.
- The heavy AI authorship is supported by the witness's own trial testimony, reported as a concession that 85 to 90 percent of the report was AI-written.
≈ What's misleading 3
- **Fabrication (of a procedural step):** the claim says "a judge ordered 3M's legal team to turn over 350 pages," and the Futurism caption says "the judge obliged." No source I found describes a judge, a motion to compel, or a court order. Every account traceable to the 404 Media reporting describes opposing counsel demanding the prompts mid-deposition, the deposition pausing for three hours, and 3M's lawyers producing the material voluntarily under discovery obligations. The gap matters because a judicial order implies the defense contested production and lost a ruling, which is a different and more dramatic event than what the transcript describes.
- **Exaggeration:** the claim says he used ChatGPT to create his "entire" report. 404 Media's own wording is "significant portions." The trial transcript figure is 85 to 90 percent, and "exclusively relied on ChatGPT" is opposing counsel's characterization, not a neutral finding. The difference between "significant portions," "85 to 90 percent," and "entire" is small in impact but real, and the claim adopts the most extreme available version.
- **Causal ordering inverted:** the claim presents the public chat links as something discovered inside logs handed over after a judge's order. The transcript quoted by Blaze indicates the public accessibility of the chats is part of how the lawyer confronted the witness in the first place, before the three-hour pause. The logs did include public links, but the sequence in the claim reads backwards.
? What's uncertain 4
- I did not retrieve the deposition transcript, the trial transcript, or the 350-page prompt log. Everything about their contents rests on named outlets reporting what those documents say, several of them quoting them verbatim and consistently.
- Whether any judge at any point issued an order touching the prompt production, for example a later ruling on admissibility or on 3M's attempt to disqualify its own witness, is not established by anything I found. Absence of a reported order is not proof no order exists anywhere in a six-year MDL.
- No confirmation, denial, or explanation from 3M, Knighthawk Engineering, or Autenrieth. All three declined to respond to the originating outlet, which is neither confirmation nor evasion.
- How much the AI report actually affected the verdict is not established. The jury found 3M 30 percent responsible, but no source ties that outcome to the expert's AI use.
Sources
6 of 8 linked to records404 Media, Jason Koebler, "'Show How 3M Is 0% at Fault:' Expert Witness Used ChatGPT to Write Report Defending Company in Deadly Explosion Lawsuit," Aug 17 2026
Above the Law, Joe Patrice, Aug 18 2026
Blaze Media, Aug 2026, quoting the deposition transcript verbatim
Plaintiffs' counsel press releases and National Law Review / PR Newswire postings on the Aug 10 2026 verdict, with cause numbers and presiding courts
ABC13 Houston and Houston Public Media coverage of the verdict, including 3M's on-the-record statement
Breitbart, Racket, The Mary Sue, Business Insurance, beSpacific, all downstream of 404 Media
Futurism, "Man Obliterated When Court Obtains His ChatGPT Transcripts"
The underlying deposition transcript, trial transcript and 350-page prompt log