Former New Jersey lieutenant governor Dale Caldwell says responses from multiple artificial-intelligence platforms support his claim that he was unfairly targeted by a sexual-harassment investigation, according to The Verge. His argument puts chatbot output at the center of an effort to dispute an official finding, despite the absence of evidence in the report that those systems conducted an independent investigation.

Caldwell resigned on September 25 after an investigation found that he had sexually harassed a staff member and repeatedly violated ethics rules, The Verge reported. He has since appeared in the media to contest the findings. During an interview on NJ PBS, host Rob Nelson asked Caldwell about his response to the report.

An investigation file and AI responses sit on opposite sides of an evidentiary boundary.
Caldwell said no AI platform he consulted reached a finding of sexual harassment.

Caldwell said he had processed the investigation report through multiple AI platforms and asked the systems what findings they would reach. He claimed that none of the AI responses concluded that sexual harassment had occurred. The Verge’s account does not identify the platforms, reproduce the prompts and complete outputs, or describe any method used to test whether the responses were reliable.

Those missing details matter because chatbot answers can change with wording, context, and the information supplied by the user. The Verge noted that AI chatbots can be highly agreeable and may reinforce what a user appears to want. On the information available in the report, Caldwell’s chatbot results cannot be independently evaluated or treated as a new factual inquiry.

The same source pages produce different AI outputs as their framing changes.
Without the prompts and complete outputs, the chatbot responses cannot be independently assessed.

The episode illustrates a growing problem as generative AI enters public disputes: a fluent answer can be presented as outside validation even when it may only reflect the material and framing given to the system. Caldwell’s account does not erase the investigation’s findings, and The Verge did not report that any authority had reversed them. The uncertainty concerns what the unidentified chatbots actually received and produced—not the existence of the reported official findings.