Urgent.News

What's breaking now, across thousands of outlets.

AI

How a ChatGPT user ‘embarrassed’ 3M in court

An expert witness for 3M extensively used ChatGPT to draft a report on a fatal Houston explosion. Prompts revealed the AI was asked to defend 3M and show zero fault for the incident. The expert acknowledged using the AI for most of the report, which became part of court discovery. This case highlights how AI conversations can be revealed during legal proceedings.

How a ChatGPT user ‘embarrassed’ 3M in court

3M engaged an expert witness to draft an expert report regarding a 2020 Houston explosion that resulted in three deaths and destroyed around 200 homes. The use of AI in the creation of the expert report became evident during court proceedings when the prompts used by Josh Autenrieth from Knighthawk Engineering were revealed. Autenrieth utilized ChatGPT to assist him in crafting a report defending 3M's standard of care and establishing 3M's 0% responsibility for the explosion at Watson Grinding.

The ChatGPT conversations, which showed Autenrieth repeatedly uploading report versions for revision and seeking the AI's assessment, were disclosed during discovery. Autenrieth testified that while the AI played a role in his process, his professional expertise formed the basis of his opinions. The jury ultimately awarded over $61 million to the plaintiffs, attributing 30% of responsibility to 3M and 70% to Watson Grinding, marking the case as one of several in which AI-generated material has surfaced in legal proceedings.

Written by urgent.news from Times of India's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

This story

This is one outlet's version. Read the fullest account.

Read the original at timesofindia.indiatimes.com →

More in AI

Pi Agent vs OpenCode after 100+ Hours of Real Use ✌️

Open-source coding agents had a weird start to 2026. In January, Anthropic suddenly blocked third-party tools from using Claude subscriptions. Overnight, with no warning. OpenCode got hit the hardest.

  • Pi Agent achieved 21/30 task success rate with 362.9 seconds median completion time
  • OpenCode completed 19/30 tasks with 280.6 seconds median completion time
  • Pi Agent edges OpenCode in efficiency, hackability, and raw results

More from Friday 21 August →