In New York, New Jersey, and Connecticut, a client wearing Meta Ray-Ban AI glasses can legally record your therapy session without saying a word about it — and there is no federal law that specifically prohibits them from doing so. The glasses ship with a small white LED that is supposed to indicate active recording, but a publicly documented $60 modification disables that light entirely, leaving the lenses indistinguishable from any other pair of glasses on your client's face. The technology is mainstream and the law has not caught up. If you do not have a written device policy in your intake paperwork, you do not have a policy.
How the glasses actually work
The Meta Ray-Ban smart glasses are the device therapists should know first. Meta launched a new $299 AI glasses model on June 23, 2026, and the prior generation remains widely available. The glasses look like standard prescription or sunglasses. They have a built-in camera, microphone, and AI assistant. When the camera is recording, a white LED on the right temple is supposed to illuminate. When you are in a session with a client wearing them, you are looking for a small light on the side of their glasses while maintaining therapeutic presence. Tom's Guide called it "the blink-and-you'll-miss-it sign that Meta glasses are recording."
That LED can be defeated. A 404 Media investigation documented a hobbyist who removes the LED with a Dremel tool and fills the housing with resin for about $60. The result records without any visible indicator. The same report noted this person had a "growing list of customers around the country." Meta's tamper protection stops the glasses from functioning if the LED is covered — but the physical removal circumvents that safeguard entirely.
The glasses are also sending data. A federal class-action lawsuit filed in the Northern District of California on March 4, 2026 (Bartone et al. v. Meta Platforms) alleged that recorded footage — including private moments — was routed to human contractors in Kenya for AI training purposes, despite Meta's marketing language about user control. The Electronic Frontier Foundation issued a warning about the glasses in March 2026, specifically flagging that bystanders have no meaningful notice when they are being recorded.
Meta Ray-Bans are the most prominent example, but not the only one. The OhO Sunshine Edge Pro is marketed with "security and evidence" language and stores up to 64GB locally without cloud dependency. The Solos AirGo 3 carries a ChatGPT integration and records audio, which is enough to capture a session without a camera at all. Google and Samsung have AI eyewear targeted for later in 2026. The category is not a niche product anymore.
The legal landscape: what NY, NJ, and CT actually say
The core legal question for a therapist is whether a client has to tell you they are recording. The answer in all three target states is no — and understanding why requires understanding what one-party consent means in practice.
One-party consent means that any party to a conversation can legally record it without notifying the other parties. In New York (N.Y. Penal Law § 250.05), New Jersey (N.J.S.A. 2A:156A-3), and Connecticut (C.G.S. § 53a-189), a person can record a conversation they are participating in without asking anyone else. A client is a party to their own therapy session. They can record it.
Connecticut has an important nuance worth knowing: the all-party consent standard in Connecticut applies to telephone and electronic wire interception, not face-to-face conversations. In your office, a Connecticut client has the same one-party right as a client in New York or New Jersey.
New York has had pending legislation — Senate Bills S5077 and S5070 in the 2025-2026 session — that would shift the state to all-party consent. As of June 2026, both bills remain in committee. New York is still a one-party consent state.
There is no federal law specifically governing in-person recording with wearable devices in a therapy context. The federal wiretapping statutes were written before glasses with built-in cameras existed. The gap between the technology and the law is real, and it is not going to close fast.
The HIPAA question — and why the answer is not what you might expect
HIPAA is probably the first thing that comes to mind here. It should not be.
HIPAA's Privacy Rule and Security Rule bind covered entities and their business associates. Your clients are neither. When a client records your session, they are not subject to HIPAA obligations, and HHS guidance is explicit about this: individuals receiving services are not bound by HIPAA when they capture information using their own devices. The agency noted this specifically in telehealth guidance, but the principle applies the same way in an in-person office. Once a client has captured information about their own healthcare encounter, HIPAA does not govern what they do with it.
What HIPAA does say — and this matters — is that you are responsible for protecting protected health information in your environment. That includes taking reasonable steps to maintain the confidentiality of what happens in your office. It does not give you legal authority over a client's glasses. The obligation runs to your systems and your practices, not to what a client chooses to do in a space you provide.
The consent and confidentiality gap here is a state-law and ethics question, not a HIPAA question. Framing it as HIPAA will not help you and may give you a false sense that a policy you already have covers this. It probably does not.
As of mid-2026, neither the APA nor NASW has issued a formal ethics opinion specifically addressing AI wearable glasses in clinical settings. The California Association of Marriage and Family Therapists noted in legal guidance that they had seen an influx of calls from therapists asking about clients recording sessions without consent — that is the closest thing to a professional organization acknowledging this as a live issue. The guidance vacuum is the system failure here. The profession has not moved fast enough, and individual clinicians are left to write their own rules.
What a written device policy should cover
The time to address recording is at intake, not in the middle of a session. A device policy embedded in your informed consent document gives you a clinical and legal foundation, makes the conversation normal rather than accusatory, and avoids having to improvise when a client walks in wearing something unfamiliar.
A practical policy should address four things:
1. Disclosure requirement. State explicitly that clients are expected to disclose if they intend to record any part of a session — audio or video, using any device, including smart glasses or recording wearables. This does not make recording illegal if they fail to disclose, but it establishes a clear expectation and creates a clinical basis for the conversation if it comes up.
2. Your position on recording. Decide whether you allow consensual recording. Some clinicians permit audio for personal review. Some do not permit any recording. Whatever your position, state it. You have the right to set conditions for your practice environment, including asking a client to remove or put away a device as a condition of continuing to meet.
3. What the client is agreeing to. If you do allow consensual recording, specify how it may be used — personal reference only, not shared publicly, not submitted for AI model training. This will not bind a client legally the way a contract would in most circumstances, but it documents the agreement and reflects the clinical frame.
4. Wearable devices specifically. Do not write "electronic devices" and assume it covers glasses. Name the category: smart glasses, AI glasses, recording wearables. Clients may not think of them as recording devices the way they think of a phone. Making the disclosure explicit removes that ambiguity.
Keeping clean, structured session documentation on your end is part of the same picture. When you have a reliable record of what was discussed, when, and how, you are in a much stronger position if a recording's accuracy or context is ever questioned — which is exactly the kind of HIPAA-eligible documentation tool that VibeCheck is built to support.
What to do if you discover mid-session a client was recording
First: do not react to the discovery as an accusation. In most states, the client did nothing illegal.
The more useful frame is clinical: what does it mean that this is happening? Is the client experiencing distrust? Are they trying to document something because they feel unsafe? Are they unaware of what it means for the therapeutic frame? The answer to those questions is more clinically relevant than the recording itself.
Practically:
Name what you observed without accusation: "I noticed you're wearing glasses with a camera. Are they recording right now?" Give the client a chance to respond before drawing conclusions.
Reference your intake agreement. If your policy covers this, you have a grounded, non-reactive position to return to. If your intake agreement is silent, this is the conversation that will prompt you to update it.
You can ask a client to stop recording as a condition of continuing the session. This is a clinical boundary, not a legal order — you cannot compel them to stop. But you can end the session if they refuse, and you can document the clinical basis for that decision.
Document what happened, when, and what was said. If the recording was covered by your intake policy, document that the policy was discussed at intake and the client signed it.
A note on AI features in those glasses
A detail worth knowing: the Meta Ray-Bans and similar devices do not just record passively. The AI assistant can identify faces and read text visible in the environment. A client wearing them in your office could, without pressing any additional button, ask the AI assistant to identify you from public photos or read text visible on your desk or wall. That is a different kind of privacy consideration than a file recording, and it does not require capturing a saved video. It happens in real time and leaves no local record.
This does not require a dramatic policy response, but it is worth factoring into how you think about what is visible in your consulting room — your license, client materials, appointment reminders, anything a camera-and-AI combination could read.
FAQ
Can a client legally record my therapy session without telling me?
In New York, New Jersey, and Connecticut, yes — all three states operate under one-party consent for in-person conversations. A client who is a party to the session can record it without notifying you. This may constitute a significant breach of the therapeutic frame, and clinically it often does, but it is not illegal. Your policy at intake is the appropriate place to address it.
Does HIPAA protect me if a client records our session?
Not in the way most therapists expect. HIPAA binds covered entities, and clients are not covered entities. HHS guidance is explicit that individuals receiving healthcare services are not subject to HIPAA obligations when they capture information on their own devices. The relevant question is state recording law, not HIPAA.
What makes AI smart glasses different from a client recording on their phone?
A phone on a coffee table is a visible, recognizable recording device. Smart glasses look like glasses. The LED indicator on Meta Ray-Bans is small enough that Tom's Guide described it as something you would miss if you blinked, and a documented $60 physical modification removes it entirely. The same device can be recording or not, and there is no reliable visual cue from across the room.
What should my informed consent say about this?
Name the category explicitly — smart glasses, AI glasses, and recording wearables, not just "electronic devices." State your position on recording. Require disclosure if a client intends to record. If you permit consensual recording, specify it is for personal reference only and may not be shared publicly or used to train AI systems. Get a signature. Revisit the language when your state's recording laws change.
---