The age of AI glasses and smart glasses is arriving.
Cameras and microphones are moving into eyewear. Artificial intelligence can process surrounding information in real time. Users can record or transmit what they see and hear almost immediately. Recording is no longer only the obvious act of taking out a smartphone and pointing it at someone. Glasses, badges, earbuds, watches and necklace-style devices can all become recording devices.
This shift is putting pressure on privacy law.
California’s SB 1130 tried to address exactly this problem. The bill would have made it unlawful to use a wearable recording device to capture another person’s sound or video in areas within a place of business where that person has a reasonable expectation of privacy, unless the person operating the device had that person’s explicit consent. The bill also would have prohibited disabling a light, sound or other indicator showing that the device was capturing sound or video.
But California Governor Gavin Newsom returned the bill without signing it on September 30, 2026.
The logic of the veto was subtle.
Newsom did not deny the problem. On the contrary, he acknowledged that wearable technology is advancing rapidly and that cameras, microphones and AI systems are now being embedded in objects that look very similar to ordinary glasses or accessories. These devices can record audio and video, process information in real time and transmit it to an audience. He also recognized that uncomfortable, invasive and potentially dangerous uses of these technologies have emerged and that a legislative response is needed.
Still, he did not sign the bill.
The reason was lack of precision.
Newsom said the issue must be approached with “precision and care.” His central concern was that SB 1130 defined several terms too broadly or imprecisely. In particular, the definition of “wearable recording device” was broad enough to include smartwatches and other worn devices that are widely used in public spaces and capable of recording. That, he argued, could create significant confusion and unintended consequences.
This is the key point.
The bill’s goal was to regulate covert recording and filming through devices such as AI glasses. But if the legal wording is too broad, the actual scope can become much larger. Smartwatches, fitness bands, action cameras, hearing-related devices, workplace wearables and medical-assistance devices may all be pulled into the debate.
In technology regulation, definitions are power.
What counts as a “wearable recording device”?
What counts as “recording”?
Which areas within a business create a reasonable expectation of privacy?
How should the law distinguish between a device that is capable of recording and an actual recording act?
Should real-time AI processing be regulated even when the system does not store the audio or video?
If these questions are not answered clearly, the law becomes vague.
And vague criminal law is dangerous. SB 1130 was not merely a recommendation or a civil compliance rule. It would have created new criminal exposure. A violation of the proposed Penal Code section could have been punished by a fine of up to $1,500, imprisonment in a county jail for up to one year, or both.
That means citizens would need to know clearly what conduct could make them criminally liable. Law enforcement would also need clear boundaries to avoid arbitrary interpretation.
That was exactly Newsom’s concern.
He agreed with the purpose. But he believed the broad definitions could sweep in ordinary devices, such as smartwatches, that people lawfully use every day. A law aimed at AI glasses could end up confusing smartwatch users.
His second reason also matters.
Newsom said SB 1130 would prohibit visual and audio recording in areas within a place of business where a person has a reasonable expectation of privacy, but California law already criminalizes certain forms of electronic recording where people have privacy expectations and also prohibits secret recording of confidential communications without consent. AP likewise noted that California is already one of the states requiring all-party consent before certain conversations can be recorded.
In other words, he did not say a new law is never needed.
He said that existing law already covers part of the problem, and adding a broad, imprecise new criminal provision on top of it could create overlap and confusion. Privacy protection may be necessary, but new crimes should not be created without clarifying how they fit with existing wiretapping, eavesdropping, secret-recording and privacy statutes.
The veto message compresses the central difficulty of regulating AI wearables.
Technology gets smaller quickly.
Law struggles to follow device names.
Today’s smart glasses may become tomorrow’s hearing aid, watch, brooch, earbud, name badge or clothing pin.
So law should be designed around conduct and risk, not merely around a device category.
SB 1130 targeted wearable recording devices.
But the more important question may not be “What were you wearing?” It may be: What did you record, whom did you notify, what context were you in and did the other person have a meaningful chance to consent or refuse?
A person can secretly record with a smartphone and violate privacy.
A person can wear smart glasses without recording and create a different legal question.
The closer technology moves to the body, the more carefully regulation must be written.
A smartphone is held in the hand. Recording is often more visible.
Smart glasses sit on the face. Wearing and recording may be harder to distinguish.
A smartwatch sits on the wrist. It may be capable of audio recording, but it is not always used as a surveillance device.
An AI badge may record a meeting for summarization, but it also raises questions about the consent of bystanders and coworkers.
Putting all of these devices into one broad legal bucket may be easy, but enforcement becomes difficult.
Newsom’s veto should not be read simply as a decision in favor of the technology industry. His message clearly recognized the risks of wearable technology. He described ordinary-looking glasses and accessories that contain cameras, microphones and AI systems, and he acknowledged their ability to process and transmit information in real time. He also recognized the emergence of invasive and potentially dangerous uses and the need for legislative response.
The problem was the method.
Newsom said he supported the need identified by the bill’s author, but the law had to be more precise. That is a basic principle of technology regulation. The emergence of a new risk does not mean that a broad prohibition is always the right answer. This is especially true when criminal penalties are involved. People need to know what is forbidden and what is permitted.
AI wearable regulation is likely to return again and again.
Smart glasses such as Meta’s Ray-Ban Meta already include cameras and microphones. As AI assistants are integrated more deeply, users may be able to analyze people, places, documents, products and conversations in real time. A device may summarize a meeting, recognize a face, translate speech, search the web, identify objects or record what the wearer is seeing.
The privacy problem divides into three layers.
First, recording.
This is the case where audio or video is stored without another person’s knowledge. Existing recording and privacy laws can address some of this, but wearables make it harder for people nearby to notice when recording is happening.
Second, real-time processing.
Even if audio or video is not stored, AI may analyze it instantly. A system could infer emotion, age, gender, identity, conversation content or document text in real time. That is a privacy problem that is related to recording, but not identical to it.
Third, transmission.
If a wearable device streams video to another person or sends data to the cloud, people in the room may not know who else is seeing or processing them. Newsom’s reference to live transmission points to this risk.
SB 1130 addressed part of this problem, but its legal definitions appear not to have tracked the technical reality precisely enough.
Good regulation should distinguish context and conduct rather than rely only on broad device names.
A restroom, changing room, medical office, counseling room, conference room, retail store, street and public park all carry different levels of privacy expectation. The law also needs to distinguish among recording, real-time analysis, storage, transmission, notice and consent.
The form of the device matters, but it is not enough.
AI glasses are not a problem simply because they are glasses.
They are a problem because recording may be hard to notice, analysis may happen in real time and bystanders may have little control.
A smartwatch is not always a problem.
But a smartwatch used as a covert recording device can become one.
Future bills are therefore likely to be built less around the broad category of “wearable recording device” and more around combinations of concealment, notice, storage, transmission, analysis and privacy-sensitive settings.
This veto shows how U.S. state governments are beginning to handle AI wearable privacy.
The first stage is recognizing the problem.
Wearable AI devices are different from older recording practices.
The second stage is legal definition.
Legislators must decide which devices and which acts should be regulated.
The third stage is integration with existing law.
New rules must be fitted into existing laws on eavesdropping, secret recording, nonconsensual filming and invasion of privacy.
SB 1130 was persuasive at the first stage.
It did not persuade the governor at the second and third.
The decision also offers lessons for countries outside the United States.
Korea already has laws and debates around illegal filming, secret recording, communications privacy, personal information protection, location information, portrait rights and workplace surveillance. But AI glasses and wearable recording devices blur existing boundaries. When a person wearing camera-equipped glasses enters a conference room, cafe, hospital, school, religious facility or workplace, what notice and consent should be required?
Technology is beginning to look like ordinary life.
In the past, a camera looked like a camera.
Now a camera can look like glasses.
A recorder may no longer look like a recorder.
AI may operate outside the screen.
This shift can weaken social trust.
When people speak with someone, they assume the other person is looking at them. They do not necessarily assume they are being recorded, analyzed and transmitted elsewhere. Wearable AI changes that basic premise. For that reason, regulation is necessary.
But regulation must be accurate.
Newsom’s veto message is about that balance. He acknowledged the risk. He acknowledged the need for legislative action. But he refused to sign a bill that, in his view, defined key terms too broadly and imprecisely, risked sweeping in everyday devices such as smartwatches and overlapped with existing privacy law.
This is a classic dilemma of AI-era legislation.
Move too slowly, and real harms go unaddressed.
Move too quickly, and the law overreaches.
Define too narrowly, and loopholes remain.
Define too broadly, and ordinary life becomes criminalized.
The veto of SB 1130 shows that tension.
The problems created by AI glasses and wearable recording devices are real. But solving them requires law that targets the actual conduct: recording, transmitting and analyzing people in contexts where they reasonably expect privacy.
It is not enough to write a broad law around the word “wearable.”
This decision does not mean wearable AI regulation is over.
It means a more precise bill is needed.
Technology has already entered the frame of ordinary glasses.
Now law must redraw the boundaries among lenses, microphones, AI models, cloud transmission, notice, consent, public space and private space.
The real question is not simply whether someone is wearing a device capable of recording.
The real questions are these:
Could people know they were being recorded or analyzed?
Could they refuse?
Was the space one where privacy could reasonably be expected?
And does the law describe those differences clearly enough?
Gavin Newsom’s veto message answered the last question this way:
Not yet.
