Long considered futuristic gadgets or commercial flops (think Google Glass or Snap Spectacles), smart glasses are making a comeback thanks to dramatic advances in artificial intelligence. Worn as part of everyday life, they no longer just record or stream: they understand, respond, translate, and interpret—all in real time.
It is against this backdrop that Meta, in partnership with Ray-Ban, has launched the second generation of its smart connected glasses. More discreet, more powerful, and above all equipped with a built-in AI assistant, they represent a new milestone in the convergence of everyday objects and automated cognitive capabilities. Far from being mere cameras mounted on temples, the Ray-Ban Meta 2025 is a true portable AI terminal, capable of interacting with the environment, processing voice commands, and providing context-aware responses.
A technical data sheet for embedded AI
Behind the familiar look of classic sunglasses lies cutting-edge technology. The second generation of Ray-Ban Meta features:
- A 12-megapixel camera capable of recording video in 1080p at 60 frames per second,
- Five built-in microphones to pick up both the user's voice and ambient sounds,
- 32 GB of storage, enough to save up to 100 videos or 500 photos,
- Miniature speakers built into the arms, allowing you to listen without headphones,
- An average battery life of 4 hours of active use, with a charging case.
But most importantly, they are the first to integrate Meta’s AI voice assistant—based on Llama 3—directly into the glasses. This assistant can answer questions, analyze a visual scene, summarize a conversation, translate in real time, and even comment on what the user is seeing. The promise? A generative, contextual, and multimodal AI, always within sight (and earshot).
Key figures: a rapidly growing market
Wearable connected devices, which include smart glasses, represent a rapidly growing market segment. According to IDC, more than 120 million units were sold worldwide in 2024 (including smartwatches, earbuds, and smart glasses). Meta aims to rapidly expand its Ray-Ban Meta lineup, with 1 million units sold for the second generation, set to launch in the spring of 20251.
The launch price in France is €369 for the base model, with options available featuring prescription lenses or photochromic lenses. The primary use remains photo and video capture (80% of users), but Meta notes a 45% increase in voice interactions between 2024 and 20252, a sign that the “conversational AI” aspect is gaining traction.
These figures reflect a fundamental trend: AI is becoming increasingly integrated into the human body, in forms that are ever more discreet, mobile, and personal.
A real-time human-machine interface
The experience offered by Ray-Ban Meta is based on the interaction between voice, vision, and intent. Users can activate the voice assistant simply by saying “Hey Meta,” then make a natural request such as:
- “Take a picture and send it to Camille,”
- “What does this word mean in Italian?”
- “What are the opening hours of this museum?”
- “Summarize for me what I told Jean earlier.”
The glasses then use a combination of visual data captured by the camera, contextual information (geolocation, time, history), and voice input to generate a relevant response.
Meta is also experimenting with more advanced features, such as object recognition, pedestrian navigation assistance (“follow that person in the crowd”), and interaction history that allows users to revisit past conversations.
What are the ethical limits of AI worn on the face?
While the innovation is undeniable, it raises numerous ethical, social, and legal questions. Wearing an AI device on one’s nose in a public place raises sensitive issues:
- Implied consent: Should you let those around you know that you’re wearing glasses capable of recording or analyzing speech?
- Protection of third parties: Although facial recognition is disabled by default, users can re-enable it. What safeguards are in place?
- Privacy and geolocation: Is the data processed locally or sent to the cloud? Meta claims that most of the processing takes place locally, but legal uncertainties remain.
- Possible misuse: covert recording, creation of false visual evidence, audio deepfakes using voice modeling…
In France, the CNIL is closely monitoring developments in this area. In May 2025, it reiterated that any recording in public spaces must comply with the GDPR, even if carried out by a private individual using a smart device3.
Is portable AI becoming commonplace?
The Ray-Ban Meta is part of a broader trend of wearable smart interfaces, alongside the Rabbit R1, the Humane AI Pin, and the Apple Vision Pro. Unlike smartphones or smartwatches, these devices combine AI with eye tracking and voice control, creating a new form of “augmented” interaction.
For now, its use remains limited to early adopters and recreational, professional, or tourism-related contexts. But as the performance of voice and visual AI improves and interfaces become more user-friendly, the widespread adoption of facial AI is becoming a plausible prospect.
The Ray-Ban Meta glasses thus raise a fundamental question: do we want our most everyday tools to become intelligent agents, capable of perceiving, interpreting, and even anticipating our actions and intentions?
Learn more
In a previous post on this blog, we examined Meta’s technical ambitions through the lens of VivaTech 2025: Mistral AI unveils a sovereign high-performance computing infrastructure in partnership with Nvidia, as well as the rise of embedded AI in “Magistral: Mistral’s artificial intelligence that brings new meaning to automated reasoning.”
References
1. IDC. (2024). Worldwide Quarterly Wearable Device Tracker.
https://www.idc.com/
2. Meta. (2025). Ray-Ban Meta Usage Data – April 2025 Report.
https://about.meta.com/
3. CNIL. (2025). Wearable Devices and Privacy: Updated Recommendations.
https://www.cnil.fr/
