New AI wearable pushes the boundaries of privacy and eavesdropping laws

Chris Thomas is all about AI gadgets.

Thomas (43 years old) metacamera glasses, a Field Day pendant for taking notes around your neck, and a wristband. AmazonBee, all records his daily life. Thomas, a Pittsburgh-based business owner and live streamer, uses his devices to stay organized, pay his bills on time, and set reminders.

“I’m very forgetful,” Thomas said in an interview. “My life is so crazy that it’s nice not to have to be reminded of it.”

Not everyone agrees with his lifestyle. Earlier this week, a bystander at a local bar noticed his Metagras.

“He was like, ‘I don’t feel well,'” Thomas said. “I don’t turn it off, I turn it off completely, because people don’t believe it’s off.”

This is the type of interaction users increasingly navigate as more tech companies introduce AI-powered always-listening devices and features, marketed as useful tools to transcribe meetings, track conversations, set reminders, and provide personalized recommendations.

The latest examples of these gadgets debuted Thursday in Amazon’s Be Frontier wearable device and Be Eclipse, a card-like device that attaches to your smartphone. Both devices are designed to record, transcribe, and analyze a user’s day.

Apple’s latest Apple Watch Series 12, announced in September, will come with Siri Recap, which takes notes “ambiently” throughout the day and generates summaries of conversations. Another feature called Live Rewind can summarize 15 seconds of conversation through text transcription.

Meta’s popular smart glasses can take photos, capture videos, and record voice memos and reminders. Startups like Plaud and Pocket are also developing card-sized note-taking devices, and Friend offers an AI necklace.

These devices allow for covert recording without bystander consent and can collect sensitive data that may be accessible to others, raising concerns that they may violate privacy.

Amazon, Google and other makers of voice-activated smart speakers faced similar privacy concerns when their devices first came to market, leading in some cases to lawsuits and regulatory fines.

But this new generation of AI devices is also pushing the limits of privacy laws.

Unlike previous smart speakers that are activated by a so-called “wake word” and in some cases save recordings or send audio to the cloud, some of these products can continuously listen to audio, interpret it, and extract its meaning without retaining the underlying audio. This feature has raised little-tested questions about whether the process itself constitutes unlawful interception.

“My statement is that new technologies like this should always be approached with caution,” said Karen Panetta, dean of graduate education at Tufts School of Engineering and a fellow of the Institute of Electrical and Electronics Engineers.

She added: “If I give permission to record myself and everything I see, that doesn’t mean everyone else around you has given that consent.”

Apple and Amazon appear to have avoided some of the privacy concerns that plagued previous voice assistants.

Apple says the feature doesn’t require storing audio, sending recordings to a server, or human review. Amazon said it does not store audio from Bee devices and that conversations are processed in a secure environment where “no one, including Amazon, has access to the data.”

However, that may not resolve the legal issue.

Many states’ wiretapping and eavesdropping laws are built around whether private communications were intercepted and their contents obtained, rather than simply whether a traditional audio recording was created.

Colin Zick, a privacy and data security attorney at Foley Hogue, said Apple appears to have learned from the first wave of lawsuits surrounding voice assistants. However, this new technology raises questions that never needed to be answered in previous cases. In other words, does a system that continuously listens to audio, interprets it, and converts it into a summary constitute eavesdropping, even if it does not store any of the underlying audio?

Some state laws define interception broadly. For example, in Massachusetts, interception device is defined to include equipment that can “send, receive, amplify, or record” communications, and the “substance” of a communication may include its content and meaning.

In Zick’s view, a device that continuously receives audio and generates a written summary could qualify as a language, even if it doesn’t hold an audio file.

And in some states, greater legal risk could fall on the person wearing the device rather than the AI ​​gadget maker like Apple or Amazon.

“Non-wearers did not purchase the watch, agree to Apple’s terms, or choose to have their conversations summarized in the app,” Zick said.

Daniel Pi, an associate professor at the University of New Hampshire Franklin Pierce School of Law, said he believes New Hampshire’s statute is sufficiently broad that the Watch’s new audio intelligence feature constitutes eavesdropping. The law covers not only recordings but also “audio or other acquisition” of the content of communications.

That could create an unusual situation in which consumers could, at least technically, violate state law by turning on features built into mainstream devices and using them during private conversations without everyone’s consent.

Pai said users in New Hampshire could face civil or criminal liability if they did so, but emphasized that the actual risk of prosecution is much lower. He said it can be difficult to know whether someone had the feature enabled in the first place, and he would be surprised to see prosecutors pursue cases involving naked violations where no one was harmed.

The legal issues are also different from those posed by previous smart speakers and voice assistants. These cases often centered on devices accidentally activating or capturing audio that was not intended as a command. Here, this feature is designed to capture and interpret the content of the conversation, and the user has intentionally opted in.

Pai said that could change the analysis and shift the blame onto the wearer rather than the technology company itself.

Broader privacy concerns extend beyond legal liability.

Michael Stokes, a former CIA official and senior vice president of emerging technologies at Vaillant, said the bigger change is that people can now be caught up in AI surveillance without ever owning or intentionally operating the equipment doing the collection.

“I think what I’m really concerned about is the bystander problem, where you can be monitored even if you don’t own the technology,” Stokes said.

He pointed to the increasing number of cameras, microphones and sensors being integrated into consumer devices, from smart glasses to watches to AI recorders. The difference now, he said, is that AI can quickly interpret what these devices collect and draw meaning from far more information than previous generations of consumer technology.

For Stokes, that makes the Apple Watch part of a larger change in the privacy landscape. While at the CIA, he helped develop the concept of ubiquitous technological surveillance, describing a world in which everyday devices continually generate information that reveals identity, location, and relevance.

The concern, he said, is not just that more data can be collected, but that the cost of understanding that data has dropped dramatically.

Stokes said it used to take thousands of hours of work to collect, categorize and analyze information about individuals. AI is increasingly automating that process, making it cheap and easy to extract useful information from data that might have been collected accidentally.

This creates new kinds of exposure for people who didn’t choose the technology in the first place, whether it’s private conversations picked up by a nearby wearable or information collected on a device they don’t own.

“We no longer have the little pockets of privacy that we thought we had before,” Stokes said.

Aida Baradari, founder of startup Deveillance, is trying to restore some control to people worried about being recorded without their consent.

The company has developed a hockey puck-shaped gadget called Specter I. The gadget uses AI to scan for nearby devices and send out a “hostile signal” that scrambles a person’s voice before the microphone can process it.

Baradari said he saw more AI wearables coming to market and felt that users should have a greater say in how they interact with them.

“We need to be able to use technology without giving up our core values ​​and rights,” she says.