Wiretap Suits Pit Old Privacy Laws Against New AI Technology

A growing stream of lawsuits challenging AI-powered customer service systems as illegal wiretaps is forcing courts to grapple for the first time with whether and how to apply older privacy laws to emerging technology.

The “conversation intelligence” technology—deployed by providers like Google LLC, Invoca Inc., and ConverseNow Technologies Inc.—can help live agents interact with customers by feeding suggestions and responses to objections, or it can replace humans altogether.

In either case, litigants say the AI system is “listening in” on customer calls in violation of federal and state wiretapping laws. At least five complaints have been filed in the US District Court for the Northern District of California so far this year, and at least a dozen since 2023.

The lawsuits’ success will depend in part on how the courts evaluate technical legal issues under the federal Wiretap Act, the California Invasion of Privacy Act, and similar laws.

Unanswered questions include whether an AI system is a “third party” to a consumer’s conversation and whether an AI tool’s nearly instantaneous interaction with a communication occurs while that communication is “in transit.”

There’s also the element of how the courts evaluate the larger implications—and possible dangers—of AI tools in general, said Matthew Pearson, a partner with Womble Bond Dickinson who defends companies in online-privacy cases and writes frequently on new developments in privacy law.

There’s a public perception that AI is going to “replace everybody” and end up “running everything,” which is far from the truth, he said.

“But I would be foolish to walk in front of a judge and say, look, it sounds nefarious, but it’s really not,” he said. “Because the court’s not going to believe me, right?”

‘Lightning Rod’

Wiretapping laws generally prohibit the recording of telephone or other communications without the consent of at least one party involved, and are often limited in scope to the interception of communications in real time.

Although wiretapping claims are nothing new in consumer-privacy litigation, the AI-focused suits are “an augmentation of existing theories that have been tried by plaintiffs’ attorneys in cases involving earlier technologies, such as chat tools, pixel tracking tools, and assistive technology to enhance the customer-service call experience,” said Melanie Conroy, a partner with Pierce Atwood LLP who focuses on privacy and data security.

At the heart of the litigation is the claim that companies using these tools have disclosed consumers’ data and communications to third parties—the providers of the tools and technologies—without informing consumers or obtaining their consent.

The use of AI technology “adds another layer,” Conroy said. AI “has become a lightning rod in our culture,” eliciting different reactions from different consumers, “and certainly could elicit different consumer decisions.”

W. Kyle Tayman, a partner in Goodwin Proctor LLP’s privacy and cybersecurity practice, sees AI as “just another technological tool that the plaintiffs will target.”

Consumers suing over data collection and disclosure through AI technology “were going to make that claim even if it was another technology rather than AI,” he said.

But the details of how AI works—and how it differs from previous technologies—could have a significant impact on the customer-call system lawsuits, Pearson said.

Previous litigation involving purportedly illegal data disclosures have involved the transfer of metadata, which isn’t easily deciphered by the ordinary person, he said.

“In a call recording case with AI listening in, it could look to a judge a lot more like what you would normally think of as a wiretap,” he said.

And the AI tool is at least capable of using a consumer’s communication to train itself and improve the service it offers, a separate use of the communication that could make it look more like a third party, he said.

Transit or Storage

Another technical issue in wiretapping cases is whether a communication was obtained while “in transit” or accessed after being stored.

Defendants in cases involving earlier technologies typically argued that although the technology was gathering and storing the communication, no one was listening in or reading it while it was taking place, Pearson said.

AI customer-call systems don’t fit that mold, he said.

“If the AI is listening, interpreting, responding, and making suggestions to the call recipient while they’re on the phone, you’re going to be closer to an ‘in-transit’ kind of interception,” he said.

Deciding whether or how the wiretapping statutes apply to AI technology will require “a deep dive into what’s happening when the technology is operating,” Pearson said.

And it will be a big lift for the courts to get up to speed on a rapidly evolving area of technology, said Calli Schroeder, lead of the AI/Human Rights Project at the Electronic Privacy Information Center.

She’s concerned that judges will defer too much to the developers of AI technology to explain how it works, an issue that’s arisen before.

Judges tend to assume the developer understands the technology best, “even when the premise of the lawsuit is that they’re not controlling it well, didn’t build it well, or aren’t giving an honest explanation of what is happening,” she said.

But judges also have a reason for skepticism, Pearson said, citing recent notable examples of AI “hallucinations” that have landed some attorneys in hot water.

“The courts are really, really, struggling with AI, and I don’t blame them,” Pearson said. “The technology is moving so fast, and the advances are coming and coming and coming, so we really need to elevate the understanding of what it is that’s happening and how it’s happening.”

“And the only way we’re going to do that is to really start litigating these things,” he said.