Michael Cohen Admits Unwittingly Using AI-Generated Fake Legal Cases

Word count: 826 Estimated reading time: 4 minutes

When Generative AI Meets the Legal System

Michael Cohen, former personal attorney to Donald Trump, recently admitted to an unusual use of artificial intelligence that landed him in hot water. In a court filing for an ongoing case, Cohen's lawyer apparently included fake legal citations generated by AI systems without realizing they were fabricated.

The incident surfaced on December 29, 2023 when a federal judge in New York directed Cohen's attorney to substantiate suspect legal citations included in a recent motion. Unsealed court filings soon revealed the citations were fakes produced by AI tools like Google's Bard and ChatGPT.

Cohen confessed to having supplied the bogus citations to his lawyer without understanding their origin. Earlier in December, Cohen had asked the AI chatbots to generate examples of prior court rulings that would support his request to terminate his existing probation early under good behavior.

According to Cohen, the AI text "looked real" to him, and he forwarded the fictional case law to his attorney without verifying their authenticity. His lawyer then incorporated the citations into a legal filing unaware they were counterfeit.

When the false citations came to light, Cohen expressed regret for the "honest mistake" and emphasized his lawyer was unaware the AI-created content was illegitimate. However, legal experts said the attorney still bore responsibility to independently validate the factual accuracy of submissions making it into official court documents.

The judge ordered Cohen's attorney to provide actual legal precedents supporting the motion in question or face potential court sanctions for the falsified filing. The attorney insisted it was merely an oversight, and stressed that the overall substance of the legal argument remained valid regardless of the botched citations.

Nonetheless, the high-profile incident highlighted the risks of AI tools like Bard and ChatGPT producing seemingly realistic but fabricated content. It re-emphasized the need for human oversight and fact checking of AI outputs before use in sensitive contexts like legal filings.

While Cohen claimed unfamiliarity with the latest AI capabilities led to trusting Bard and ChatGPT's word, experts noted legal professionals have an obligation to understand the technology tools impacting their field. Overall, the episode served as a stark reminder that generative AI remains imperfect and using outputs irresponsibly can have consequences.

FAQ

Q: What exactly happened in this case?

A: Michael Cohen submitted fabricated legal citations generated by AI chatbots to his lawyer without realizing they were fake. His lawyer then included the fake citations in an official legal filing without properly vetting them. When the judge discovered the citations were illegitimate, it caused issues for both Cohen and his attorney.

Q: Why is using AI-generated legal citations problematic?

A: Legal filings and citations need to be factually accurate and verifiable. AI systems are still imperfect and can produce falsified or fictional information that looks legitimate on the surface. Relying on unvetted AI outputs risks spreading misinformation, especially in sensitive contexts like court documents.

Q: Who is ultimately responsible when AI outputs are inaccurately used?

A: Although Cohen claimed unfamiliarity with AI capabilities, legal experts say his attorney still bore primary responsibility to independently validate any factual claims or citations included in official filings. Professionals using AI must understand its limitations and not assume outputs are inherently reliable without verification.

Q: What can be done to prevent similar issues going forward?

A: Increased education about AI capabilities and limitations is important, especially for legal and other sensitive fields. Outputs also need human oversight - people shouldn't fully trust AI responses without critically evaluating accuracy. Fact-checking AI claims is crucial before using them in impactful real-world contexts.

Q: Do generative AI systems pose new risks to the legal system?

A: As generative AI advances, there are concerns about its potential misuse to deceive or spread misinformation more convincingly. The legal community will need processes and technologies to identify and mitigate issues like AI-falsified evidence or documents entering the system. Increased scrutiny and validation of AI-assisted legal work products may be required.

Sources

[1] Michael Cohen says he unwittingly sent AI-generated fake legal cases to his attorney

[2] Michael Cohen admits to using fake AI-generated court cases in legal filing

[3] Ex-Trump fixer Michael Cohen says AI created fake cases in court filing

[4] Michael Cohen used fake cases created by AI in bid to end his probation

[5] Michael Cohen gave lawyer fraudulent case citations generated by AI