Meta Developing Prototype of Always-On AI Glasses with "Super Sensing" Capabilities

Meta is venturing into the development of always-on AI wearables. The company is currently prototyping smart glasses, dubbed “super sensing,” that can continuously record audio and capture photos every few seconds. This development was reported by the Financial Times, citing multiple sources.

When wearing these glasses, users can ask Meta AI questions about the captured audio or images. For instance, they could review a comment they missed during a meeting or instantly get information about an object they saw on the street. However, Meta is reportedly exploring a system where the raw audio and video data would not be stored, either by users or by Meta itself.

The System for Non-Storage of Raw Data

According to the Financial Times, one proposed system ensures that the raw audio and video data is neither stored by Meta nor accessible to the user. Instead, only metadata extracted from these inputs would be sent to Meta’s servers, where the company’s AI would process user queries. Insiders argue that this system minimizes privacy impacts.

Specifically, the glasses would continuously convert surrounding audio into text and extract information such as objects, text, and landmarks from images. This metadata would be temporarily retained, allowing users to ask questions like, “What was said about [X] in the earlier conversation?” The AI would then search the relevant metadata to provide an answer. The absence of raw data storage is posited as a way to reduce the risks of privacy violations.

Growing Privacy Concerns

If such “always-aware” functionality were to be implemented, it would undoubtedly raise serious privacy concerns. Meta has already faced multiple privacy-related challenges.

Firstly, the company has been under intense scrutiny regarding its use of facial recognition technology. Strict regulations in regions like the European Union and certain U.S. states have thus far prevented Meta from including facial recognition in its Ray-Ban Stories smart glasses. However, internal reports suggest that technical preparations for such functionality are ongoing.

Secondly, incidents involving users of existing Ray-Ban Meta smart glasses capturing images of individuals without their consent have sparked widespread criticism. Always-on recording in public spaces poses a significant risk to the privacy of passersby.

Thirdly, third-party modification services have emerged, offering to remove the LED recording indicator on these glasses for a fee. In response, Meta announced an update on July 8, 2026, that would automatically disable the camera if the LED indicator was tampered with.

The Controversy Surrounding LED Indicators

The handling of the LED indicator in “super sensing” mode has become a major point of contention. According to the Financial Times, Meta plans to keep the LED indicator off during this mode.

In a white paper released in July 2025, Meta explained that the LED indicator would only light up during “active capture,” such as when users explicitly save photos or videos. It would remain off during AI functionalities like reading menus. The company argued that constantly lit indicators might desensitize users, making it harder to differentiate between times when actual recording is happening.

This policy is expected to draw strong criticism. In “super sensing” mode, the glasses constantly “sense” their surroundings, yet people nearby would have no visual indication that this is happening. Meta’s spokesperson Dave Arnold told The Verge, “We do not comment on internal prototypes, but we aim to create products that are loved by both wearers and the people around them,” emphasizing the company’s commitment to “privacy by design” in its technology development.

Potential Use in AI Training

Meta is also debating whether the metadata collected through the super sensing feature could be used to train its AI models. The company has previously utilized publicly shared user photos and captions for AI training, and the real-time data from smart glasses could offer richer contextual learning opportunities.

However, how the no-raw-data-storage policy will apply to AI training remains unclear. It is uncertain whether useful training can be conducted solely with metadata or if anonymized raw data could be stored separately for this purpose.

Potential Expansion to Existing Products

The Financial Times reported that Meta might extend the super sensing functionality to its existing Ray-Ban Meta smart glasses through future updates. If this becomes a reality, hundreds of thousands of current users could gain access to always-on recognition features through a software update.

However, it remains uncertain whether the battery life and processing capacity of current devices could support continuous recording, photo capturing, and AI querying. The existing Ray-Ban Meta glasses offer around four hours of usage on a single charge, and the additional processing demands could significantly reduce this duration. Meta is also reportedly developing next-generation glasses with larger batteries.

Comparison with Competitors

Meta’s move is likely to accelerate competition in the AI wearables market. Other products, such as Humane’s AI Pin and Rabbit R1, either lack always-on recording capabilities or implement them in a limited manner. Speculations about Apple’s augmented reality glasses, “Apple Glass,” have yet to include any mention of always-on recording features.

Meanwhile, startups like Plai Labs and Brilliant Labs are focusing on AI-powered eyewear and aiming to introduce always-on recognition capabilities. Considering Meta’s scale, AI expertise, and manufacturing and distribution channels through its partnership with Ray-Ban, the company appears to have a significant advantage in this emerging field.

Regulatory Trends and Future Challenges

Regulations for wearables capable of always-on recording are tightening, particularly in the European Union. The EU’s General Data Protection Regulation (GDPR) requires explicit consent for the collection of personal data, which could make always-on recording without consent a violation. In the U.S., the California Consumer Privacy Act and Illinois’s Biometric Information Privacy Act also impose strict limits on the collection of facial recognition and biometric data.

Whether Meta’s proposed “no raw data storage, metadata-only” system can comply with these regulations will depend on legal interpretations. Even metadata could be classified as personal data if it allows for inferences about conversations or behavior patterns.

Another major hurdle is public acceptance. A society where everyone is constantly being recorded may inhibit free speech and undermine social trust. Concerns about a “surveillance society” are likely to fuel opposition from civil rights groups and privacy advocates.

Having already faced criticism over facial recognition and always-on recording with its Ray-Ban Stories, Meta appears to be proceeding cautiously with the announcement of its super sensing feature. However, given the company’s reliance on data collection and AI, striking a balance between privacy protection and functional expansion will be crucial to the product’s success.

Editorial Opinion

In the short term, Meta’s efforts could ignite new demand in the smart glasses market. The always-on recognition feature may provide significant added value, particularly for business applications like meeting transcription and real-time information retrieval. However, if privacy regulations tighten or critical media coverage intensifies, the product’s launch could face delays or functional limitations. Whether a specific product announcement will occur within the next three to six months remains uncertain.

From a long-term perspective, society needs to have a serious discussion about the implications of normalizing always-on recording and surveillance. How will the routine recording of public behavior impact freedom of expression and societal trust? As not only Meta but also companies like Apple, Google, and smaller startups pursue similar functionalities, the ethical standards of the entire industry will come under scrutiny.

The editorial team believes that a more transparent discussion about balancing technological potential with social risks is essential. If Meta is serious about its commitment to “privacy by design,” it should disclose specific technical details and establish frameworks for third-party verification of its data handling practices.

References

• The Verge: Meta is reportedly working on smart glasses that would be recording all the time — Published on July 8, 2026