By Michael Scott
Cohen’s legal team must demonstrate the presence of cases they mentioned as speculation spins around possible artificial intelligence-created content.
Court Orders Copies of Legal Citations
Attorneys for Michael Cohen, Trump’s former attorney, have been told to hand in copies of three legal cases highlighted in their motion to end his supervised release. According to the Court, the issue is that these cases do not exist. This has resulted in some people speculating that artificial intelligence (AI) was utilized to create the presented citations or motion.
In a Tuesday court filing, a federal judge cautioned that all submissions are affirmed declarations. According to the Court’s motion, Cohen’s legal team should produce a ‘detailed account’ of how the motion came to mention nonexistent cases. The explanation must also mention ‘Mr. Cohen’s role, if any, in evaluating or drawing up the motion before its filing.’
Jesse Furman, a U.S. District Judge, wrote that before December 19, 2023, Cohen’s legal team should have produced copies of the three mentioned decisions. Suppose this does not happen by this date. In that case, Mr. Schwartz should defend himself in writing by stating why sanctions, according to Rule 11 of the Federal Rules of Civil Procedure of 1927 and the Court’s inherent power for mentioning nonexistent cases, should not be applied to him.
AI Errors Halts Cohen Plans to End Court Supervision
In 2018, Cohen was handed a three-year imprisonment for wrongdoings, which included tax evasion, campaign finance contraventions, and lying to Congress. Furman claimed that decisions concerning Cohen’s motion would remain in place until his attorneys could offer evidence of the cases’ existence. This occurrence is not the Court’s first challenge in Manhattan.
On Twitter, Michael Lee, an author and journalist, said that Cohen’s attorneys or his artificial intelligence might have cited fictional cases to terminate supervised release early on. Inner City Press has been reporting this matter, similar to how it did concerning SDNY’s misutilization of artificial intelligence.
Despite the New York court not explicitly mentioning artificial intelligence in the filing, other legal cases that evoked the same question have already come up this year. Some months ago, Steven Swartz, an attorney in Mata v. Avianca Airlines, confessed that he ‘utilized’ ChatGPT for research and incorporated the artificial intelligence’s responses in court documents. The issue is that AI hallucinations fabricated ChatGPT’s findings.
In June, Mark Walters, a radio host based in Georgia, filed a case against OpenAI, ChatGPT’s creator. This happened after being accused of misappropriation by the chatbot in The Second Amendment Foundation v. Robert Ferguson.
OpenAI’s ChatGPT Lies Damaging Client’s Reputation
Mark Walter’s lawyer, John Monroe claimed that OpenAI destroyed his client’s reputation and generated shocking lies regarding him. He added that the only option was to file a case against the artificial intelligence developer.
Two months ago, lawyers for Pras Michel, an ex-Fugees member, asked for a new trial, asserting that his previous legal team utilized AI and the model hallucinated its replies. This resulted in Michael losing the case.
On Twitter, Elizabeth Wharton, Silver Key Strategies’ founder and attorney, cautioned lawyers to stop trusting artificial intelligence. They should cease using artificial intelligence and ChatGPT without confirming the information.
OpenAI has improved efforts to address hallucinations by artificial intelligence, to the extent of recruiting the alleged red teams to evaluate its artificial intelligence models for susceptibilities and gaps.
Fetch AI Partner with SingularityNET to Resolve AI Hallucinations
On Thursday, SingularityNET and Fetch AI revealed a collaboration to ultimately address artificial intelligence hallucinations and the technology’s tendency to use decentralized technology to produce incorrect or irrelevant outputs.
Alexey Potapov, SingularityNET’s Chief AGI Officer, told a media outlet that SingularityNET has been working on several methods to eliminate hallucinations in large language models (LLMs).
Neural-symbolic integration is the primary theme across all these. He also claimed that, according to them, LLMs cannot move so far in their present form and are inadequate to take them towards artificial general intelligence. However, they are a possible disruption from the final objective.