Start here
The free tool below runs entirely in this browser — no account, nothing stored or transmitted. Redact anything you would not want kept before you type it.
Open the companion tool →Why capture scope is the right question
A wearable assistant that waits for you to press a button is a different device from one that listens continuously for a wake phrase, or one that records through the day so it can 'remember' things for you. The first captures what you choose; the second captures everything in range, including people who never agreed to be recorded. Most of that behavior is configurable, and most owners never open the settings.
Where the capture goes matters as much as what is captured. On-device processing, the vendor's own cloud, and a third-party model provider are three different exposures with three different retention rules — and whether your voice or images are used to train or improve a model is usually stated in the privacy policy rather than anywhere in the device's menus. Those answers are findable in about ten minutes, and they are the difference between a device you control and one you wear.
The protocol
- Find out when it captures: is there a wake word, an always-on microphone, a camera that records on command, or continuous capture for a memory feature? Look for an indicator light and check whether it can be disabled.
- Read the retention answer: how long recordings or transcripts are kept, whether you can delete them, and whether deletion works per item or only account-wide.
- Check where processing happens — on the device, in the vendor's cloud, or through a third-party model provider — and whether the policy names that provider.
- Find the training setting: whether your voice, images, or interactions improve or train models, and how to opt out.
- Consider everyone else inside the arc: family, coworkers, patients, clients, students, and strangers. Recording other people is regulated differently by state and country, and some settings — a clinic, a workplace, a classroom — need a rule rather than a toggle.
- Change what you found, write down what you changed and when, and re-check after any firmware or app update, because these settings move.
What the tool does
Audit what your AI wearable captures: when it listens, what it records, where that goes, how long it is kept, and who else is inside its arc. It runs in this browser; nothing is sent or stored.
Limit first: This is a settings and scope audit, not legal advice, and it cannot confirm what a vendor actually does with your data. Recording other people is regulated differently by state and country, and this tool does not tell you whether a specific recording was lawful.
Open the companion tool →One human next step
If the device captured something it should not have — a medical appointment, a workplace meeting, a classroom, or someone who objected — delete it, tell the people involved, and ask the vendor in writing whether a copy exists on its servers. For a workplace or school device, raise it with the privacy or IT contact before the next use; complaints about a company's data practices go to the FTC and to your state consumer protection office.
Research log and safety checks
AI-assist path: not applicable. The audit reads the vendor's own documented behavior and the device's own settings; a model's summary of a privacy policy would add an unverifiable layer exactly where the reader needs the precise setting name (§1.2 UIC-10). §6.9 disclosed: no reliable zero-cost in-browser assist exists for identifying a specific device's real-world data flows — the honest artifact is the settings inspection plus a written question to the vendor. §6.10: not a frontier-pace topic.
§0.5 protection result: category is data-privacy-exposure; crisis-adjacent: no; fear/urgency/scarcity toolkit used: no. The protective function is free and the paid feature is convenience only.