Court filing defending an 'anti AI' law has AI generated sections

A few days ago, the artificial intelligence segment dropped one of the most curious stories that has emerged lately. Apparently, the arguments supporting an “anti-AI” law showed signs of having been generated by AI. Now, the main defender of the law admitted to having used ChatGPT in his court filing.

The background to the challenge to Minnesota’s “anti AI” law

The law, called “Use of Deep Fake Technology to Influence An Election” and presented in Minnesota, is facing a challenge. Jeff Hancock, founding director of the Stanford Social Media Lab and disinformation expert, presented a series of arguments in favor of the legislation. To be fair, the law only targets deepfake-focused platforms, not artificial intelligence as a whole. It targets artificial intelligence services capable of producing realistic content that may cause confusion.

Christopher Khols, a conservative YouTuber, and Mary Franson, a Minnesota state representative, are driving the challenge to the law. They detected certain patterns of AI use in the citation section of Hancock’s filing. More specifically, the citation section of Hancock’s filing contains references to sources and studies that don’t exist. All evidence suggested that these fictitious citations were the result of artificial intelligence hallucinations. Therefore, the plaintiff requested the exclusion of the document from the case for being unreliable.

The court filing in favor of the law has sections generated by ChatGPT

In response, Jeff Hancock has admitted that he did, in fact, use ChatGPT in his court filing. However, he asserts that he did not utilize OpenAI’s service to draft his arguments. “I wrote and reviewed the substance of the declaration, and I stand firmly behind each of the claims made in it, all of which are supported by the most recent scholarly research in the field and reflect my opinion as an expert regarding the impact of AI technology on misinformation and its societal effects,” he stated.

Then, why did Hancock use ChatGPT? According to his latest filing, he utilized ChatGPT to enhance the citation section of the document. He claims he used the AI service “to identify articles that were likely to be relevant to the declaration so that I could merge that which I already knew with new scholarship.”

Hancock says he did not intend to mislead the court in any way and apologized “for any confusion [the use of AI] may have caused.” However, he stands by the entire content of the document. That said, Hancock has a problem, since the detection of AI in one of the sections of the filing could render the rest of the content dubious. Everything will depend on how the judge receives the explanations provided by the disinformation expert.