The Dangers of Using AI to Help Your Attorney
- Pat Pegan
- Mar 2
- 5 min read
Is Your AI Assistant Helping or Hurting Your Case?
In today’s fast-paced legal world, it is tempting to turn to a "digital assistant" to help make sense of a complex case. Whether you are using Claude, ChatGPT, or another AI tool to brainstorm arguments, summarize evidence, or research legal theories, you likely feel you are being a proactive and helpful client. Using AI to help your attorney may seem like an efficient way to stay organized, but it can create significant legal risks.
However, a groundbreaking federal court ruling has sent a shockwave through the legal community, and it serves as a dire warning for anyone involved in a legal proceeding: Your conversations with AI are not private.
In the recent case of U.S. v. Heppner, No. 25 Cr. 503 (S.D.N.Y.), a federal judge ruled for the first time that a party's interactions with an AI platform were not protected by attorney-client privilege or the work-product doctrine. While this specific case arose in a criminal context, the implications are universal. In any civil or criminal matter, your AI inputs and outputs can be a prime target for the opposing side. This decision means that the "help" you think you’re getting from AI could actually become a discoverable roadmap for your adversary to be used against you, potentially exposing weaknesses in your case, your case strategy, and more.
If you or someone you know needs legal assistance in Massachusetts, fill out our contact form or call us at (978) 744-1220 for a free case evaluation.
Understanding U.S. v. Heppner
The case involved Bradley Heppner, an executive of the publicly traded company GWG Holdings, Inc., who was facing federal charges related to an alleged $150 million fraud. After he was subpoenaed by a grand jury and became aware he was the target of a criminal investigation (but before his actual arrest), Heppner used the AI platform Claude to organize his thoughts and draft potential defense strategies. He later shared these AI-generated documents with his legal team.
When the government seized his electronic devices, they discovered 31 documents memorializing these AI interactions. Heppner’s lawyers argued that these documents should be protected by the attorney-client privilege and the work-product doctrine, as they were created specifically to assist in his legal defense.
However, U.S. District Court Judge Jed S. Rakoff ruled otherwise. The court’s decision highlighted a critical mistake: Heppner acted on his own volition, without the direct supervision or instruction of his lawyers. While the defendant may have thought he was using AI to help his attorney, the court found that because Claude is not a licensed lawyer and because the interactions were not strictly confidential, those protections simply did not exist.
When the government seized his electronic devices, they discovered these records. Heppner’s lawyers argued they were protected from disclosure. However, Judge Jed S. Rakoff of the Southern District of New York disagreed, ordering that the documents be turned over to the prosecution.
The Dangers of using AI to Help your Attorney
The judge’s decision rested on several key legal principles that every client should know before they open an AI chat window:
AI is Not a Lawyer: The attorney-client privilege only protects communications between a client and a licensed professional who owes a fiduciary duty of loyalty. Because an AI platform is not a person or a licensed attorney, the court found that any "conversation" with the tool is no different from talking to a friend or neighbor about your case.
No Expectation of Confidentiality: To keep a communication privileged, it must be kept confidential. Judge Rakoff pointed to the privacy policies of consumer-grade AI tools, which often state that the company collects data to "train" the models and reserves the right to share that data with third parties or government authorities. By clicking "agree" to these terms, you essentially waive your right to privacy.
The "Work Product" Trap: Usually, documents prepared in anticipation of litigation are protected. However, the court ruled that this protection only applies when the work is done at the direction of an attorney. Because Heppner used AI to help his attorney on his own initiative, the documents were considered "independent research" rather than a protected legal strategy.
No Retroactive Protection: You cannot make a document privileged just by sending it to your lawyer later. If a document is unprivileged at the moment it is created (like an AI chat log), it remains discoverable even after it hits your attorney’s inbox.
Lack of Legal Advice
Finally, the court looked at the purpose of the communication. While the defendant intended to use the output for his case, the AI itself explicitly disclaims providing legal advice. The judge even noted that when the government asked the AI if it could provide legal recommendations, it replied that it was not a lawyer. Because the "help" was self-directed and not done at the instruction of a human lawyer, the court refused to shield the documents from prosecutors.
Best Practices: Protecting Yourself and Your Case
If you or someone you know is involved in a legal matter, the urge to "self-help" through AI is understandable, but it is incredibly risky. To ensure your defense strategy remains confidential, especially in the age of AI, follow these actionable steps:
Ask Before You Type: Never input details about your case, your concerns, or your communications with your lawyer into any AI tool without getting explicit permission from your attorney first.
Treat AI Like a Public Forum: Imagine that anything you type into a chatbot could be read aloud in court by the opposing side. If you wouldn't say it in a public square, don't say it to an AI.
Coordinate with Counsel: If you feel that AI could be useful for your case, talk to your lawyer about it. If your attorney directs the research or uses a specialized, "closed" legal AI system, there may be a much stronger argument for protecting that work.
Don’t Delete Your AI History: Most people know not to delete emails or shred documents once a case starts. You must now add "do not run AI searches about my case" to that list of prohibited activities.
Share This Warning: If you have a friend or family member working with a lawyer, make sure they understand these risks. A single late-night session with a chatbot could accidentally hand the "keys to the kingdom" to their adversary.
Contact the Law Office of Patrick J. Regan
If you or someone you know needs legal assistance in Massachusetts, don't leave your future to chance or unsecure technology. Call Attorney Patrick J. Regan at (978) 744-1220, email him at office@regan.law, or fill out the contact form.
For more than 30 years, Attorney Patrick Regan and his team have been fighting hard to advocate for their clients' rights in criminal defense and personal injury matters. We understand the evolving landscape of modern litigation and the critical importance of protecting your privilege.