A Lieutenant Governor Resigned Over a Sexual Harassment Probe. His Defense: What He Says AI Tools Told Him
Dale Caldwell, the former lieutenant governor of New Jersey, has cited responses from several AI services as evidence of his innocence. The independent investigative report released by the state concluded that his sexual language amounted to sexual harassment.
Dale Caldwell, the former lieutenant governor of New Jersey, has pushed back against sexual harassment allegations by pointing to answers he says he got from AI. According to an October 4 report (local time) by The Verge, Caldwell said he fed the investigative report into several AI services. He then claimed that none of them concluded that his conduct amounted to sexual harassment.
What the investigators concluded
The independent investigation began with an anonymous complaint filed in May. According to the report released by the state of New Jersey, investigators examined sexual language Caldwell directed at a female employee. They concluded that it constituted sexual harassment under the state’s workplace anti-discrimination policy. Possible violations of ethics rules were also examined separately.
New Jersey Governor Sherrill released the findings on September 24 and called on Caldwell to resign, citing repeated policy violations. The Verge reported that Caldwell stepped down on September 25, following an investigation that found sexual harassment and repeated ethics violations.
Caldwell has denied the allegations. His responses included in the investigative report also state his position that the inquiry would clear his name. After resigning, he went on to repeat the same claim in press interviews. It was during that stretch that he said he had asked several AI services to review the report.
Can an AI’s answer serve as evidence?
The evidence Caldwell has offered is the responses he says he obtained after feeding the report into AI tools. He claimed that none of the services concluded there had been sexual harassment. The investigators’ determination was the result of applying state policy, while the account of what the AI said comes from Caldwell himself.
To verify the claim, you would need to see how much of the report was submitted to the AI, the exact questions asked, and the answers given. Whether his account of the AI backing him up matches the actual conversation logs is another thing worth checking. It also remains possible that he cherry-picked only the responses that favored him. In the end, this is a case in which a public official who left office over an investigation’s findings has enlisted AI answers as grounds for his defense.