The New Mexico Supreme Court↗ fined defense attorney Stephen Aarons $5,000 and held him in contempt after an appellate brief contained fabricated police testimony and witness accounts generated by ChatGPT.
The case is especially stark because it involves a murder appeal for a client serving a life sentence. The court focused on the failure to verify the output against the actual trial record.
The professional risk is not merely that a model can be wrong; it is that fluent output can look authoritative enough to bypass normal skepticism.
For businesses deploying AI agents, the same design rule applies: separate generation from authority. Let the model draft, classify and propose, but require deterministic checks or human approval before consequential actions.
A New Mexico defense attorney was fined and held in contempt after an appellate brief contained invented testimony generated by ChatGPT. The case underscores that AI can accelerate professional work but cannot replace source verification.