Last Thursday, Amazon unveiled the first new hardware from Bee, the AI wearable startup it bought in 2025. The $74.99 Bee Frontier goes on your wrist or clips to your clothes. The $99.99 Bee Eclipse snaps magnetically onto the back of your phone and is coming soon.

What the Bees do is simple to describe. You turn one on, it listens, and AI turns what it hears into meeting notes, reminders, and follow-up tasks. Amazon’s devices chief, Panos Panay, put it bluntly onstage: “They listen to your whole day.” Bee can link to Gmail, Google Calendar, and Outlook to draft follow-up emails or calendar invites from what was said, and it also connects to Alexa+ and Amazon.com.

To be fair to Amazon, the privacy engineering for the wearer is real. Audio is processed in an isolated, encrypted environment and then deleted, and Amazon says only the customer can see the results, not Amazon itself.

The problem is everyone else. Jeff Williams, a former senior security engineer at Amazon Web Services, argues Bee asks the wrong person for consent. The reminder to get permission shows up only on the wearer’s phone, and nothing enforces it. The indicator light can disappear under a sleeve or into a pocket. And your coworker’s or neighbor’s words still land in someone else’s transcript, which they can’t see or delete.

The law hasn’t caught up either. About a dozen states, including California and Washington, generally require everyone in a private conversation to agree before it’s recorded. Amazon’s defense is that no audio file is kept. But some states, like New Hampshire, define interception to include simply acquiring what was said, which leaves open whether temporary AI processing counts even when the audio is deleted. As Tufts Engineering Dean, Karen Panetta, noted…. a wearer agreeing doesn’t mean the people around them have.

This is not a secret spy device. It’s opt-in, the wearer’s data is well protected, and Amazon added an auto-shutoff when it stops hearing the owner’s voice. But it moves the default from “Alexa listens when you call her” to “your wrist listens to everyone you meet.”

The safeguards protect the buyer, not the bystander. And Amazon has rolled back privacy promises before. For example, last year Amazon ended the “Do Not Send Voice Recordings” option on some Echo devices.

How worried?: 🟡 3/5 — Meaningful

(This is a fictional scenario. None of this has happened.)

It’s 2029. Listening wearables are as common as earbuds. Nobody notices the little lights anymore.

Maria is at a crowded pharmacy counter. She quietly asks the pharmacist about a prescription. Three strangers in line are wearing listening devices. Their AI assistants dutifully log the conversation, flag a “health topic,” and tuck it into three transcripts Maria will never see.

Then a quiet policy update arrives, the kind people click “Accept” on without reading. To “improve recommendations,” summaries are now retained longer and become searchable inside a new partner program. Insurers and background-check firms sign up. Nobody is buying audio. They’re buying summaries.

Six months later, Maria applies for life insurance. Her premium comes back mysteriously high. Her job offer stalls after a “routine screening.” She never wore a device. She just stood near people who did.

In court, the company points out that it never stored audio, only text. The wearers had agreed to the terms. Maria never agreed to anything. Nobody asked her.

The scariest part of this story isn’t a rogue AI. It’s that every step along the way could seem reasonable on its own.

Why it’s (probably) not tomorrow: Today Amazon says no one, including Amazon, can access Bee data, and summaries are encrypted with keys controlled by the user’s device. For this scenario to happen, that promise would have to change. That’s exactly why it’s worth watching.