Lindsay Lieberman. (Photo by Sandra Morlet)

A few months ago, I attended a legal conference where artificial intelligence was a dominant topic of discussion. As an attorney who represents victims of intimate image abuse and gender-based violence, I was eager to hear how the profession was thinking about AI.

One packed presentation in the conference’s largest ballroom focused on AI-generated evidence. Four presenters walked attorneys through three scenarios in which women used AI to fabricate evidence against men. The scenarios were illustrated with AI-generated videos depicting the women assembling their false cases, set to ominous, crime show music. One involved a workplace sexual harassment claim; another involved a custody dispute. At one point, a presenter described the woman in the custody scenario as using AI to “destroy a man’s life.”

As the presentation went on, I grew increasingly uncomfortable with its repeated focus on women fabricating claims. Over the past 15 years, I have worked with survivors navigating workplace harassment and abuse in family, civil and criminal matters. I have seen how hard they must fight to have their evidence taken seriously. I have never encountered a survivor who used AI to manufacture it.

After the presentation, I approached one of the panelists and asked how often people were actually using AI to fabricate evidence in cases like these. He could not point to one documented case. The scenarios, he explained, were meant to illustrate what AI could do.

Hundreds of attorneys had spent one hour and 20 minutes considering scenarios in which women fabricated evidence of abuse. I would later present on the harm AI is causing women and girls—but to a smaller audience in a much smaller room.

In one of my cases involving nonconsensual intimate image dissemination, law enforcement initially told my client they could not make an arrest without an IP address linking the perpetrator to the emails, even though strong circumstantial evidence identified him as the sender. I had to persuade investigators to examine that evidence rather than treat the missing IP address as the end of the inquiry. When investigators already overlook available evidence because one digital identifier is missing, the unsupported suggestion that evidence “could be AI” risks becoming another excuse to stop investigating.

Digital evidence is already difficult for victims to explain and for the legal system to evaluate. An unsupported suggestion that evidence “could be AI” may give decision-makers one more reason to stop looking closely.

Meanwhile, people are using AI tools to create sexualized images of my clients without consent. Those images are used to humiliate, threaten and extort. Victims are left explaining to law enforcement that graphic images are fake while dealing with the very real damage caused by their circulation. The images are fake. The abuse is not. The evidence is real.

These are the problems lawyers need to know how to address now. These are not hypotheticals.

The topics we give a platform at legal conferences shape how attorneys argue cases and how judges evaluate them. If lawyers leave with the impression that fabricated evidence is a common feature of abuse cases, defense attorneys may raise unsupported claims that authentic evidence was made with AI, and judges or jurors may give those claims more weight than they deserve. A survivor could then be forced to defend the authenticity of real evidence before anyone addresses the abuse it documents.

That concern is especially acute in sexual harassment and abuse cases. Survivors already face questions about their credibility, their memories and why they came forward when they did. They should not face a new presumption of doubt simply because AI could have been used to create evidence.

Lawyers, law enforcement and judges need to understand how to evaluate digital evidence: where a file came from, what its metadata can and cannot tell us, and how to consider other evidence alongside it. That includes circumstantial evidence of who was responsible, such as motive, opportunity and conduct before or after the incident.

There is enormous demand for AI programming, and legal conference coordinators are understandably eager to meet it. But the urgency to teach about AI makes it more important to scrutinize the examples we choose and the assumptions those examples leave with an audience.

Legal conferences have a responsibility to vet their speakers, ask whether the scenarios they present reflect documented problems, and give real harms the attention they deserve. Lawyers need AI education grounded in evidence and context, without presenting gendered hypotheticals that fail to reflect reality.

Lindsay Lieberman is the founder of Lindsay Lieberman LLC, a law and consulting agency, where she represents victims of intimate image abuse and gender-based violence. A former Brooklyn, New York City, prosecutor, she has spent 15 years working with survivors across criminal and civil matters.

ABAJournal.com is accepting queries for original, thoughtful, nonpromotional articles and commentary by unpaid contributors to run in the Your Voice section. Details and submission guidelines are posted at “Your Submissions, Your Voice.”