Meta Ditches Ray-Ban, Goes Alone—And Brings Kylie Jenner Along.

F or five years, Meta hid behind a prominent name. Ray-Ban. Oakley. Two brands that shouted “cool” before the tech specs really mattered—because nobody, frankly, wanted to be caught wearing a Facebook product on their face.

That era ended Tuesday.

At a debut event in New York City, Meta unveiled three self-branded AI smart glasses: the Meta Adventurer, the Meta Fury, and a celebrity collab that nobody saw coming—Meta Glasses by Kylie, internally nicknamed Starfire. The Adventurer and Fury start at $299. The Jenner edition? $399.

For perspective, the second-gen Ray-Ban Meta currently sits at $379. The Oakley Meta starts at $499. Meta recently undercut its own premium range by up to $200. That's not a product launch. That's a statement.

Here's the thing nobody hates to admit: the hardware is almost identical to last year's Ray-Ban Meta Gen 2.

You're getting a 12-megapixel camera, 3K video, a five-microphone array, open-ear speakers, and eight hours of battery (32 with the case)—which, if that list seems similar, it should, because it's exactly what the Ray-Ban Meta Gen 2 launched with. The company's VP of wearable devices, Ming Hua, didn't even try to conceal it. She described the new line as built on "the same hardware platform as our best-selling AI glasses."

What changed is everything around the hardware. Twenty-six styling combinations across the three models. A three-way adjustable nose pad that truly fits different facial shapes – something noticeably absent from earlier editions. Over-extension hinges and adjustable temple tips. Small tweaks. Real ones, but few.

The broader software story is Muse Spark – the first AI model out of Meta Superintelligence Labs, the division helmed by Alexandr Wang. It's also Meta's first closed-weight model, which is a major break from the open-source Llama policy that created the company's AI credibility. Muse Spark runs in three modes: Instant, Thinking, and Contemplating—exchanging response speed for reasoning depth based on the query. Meta says it matches Llama 4 Maverick's performance at one-tenth the compute cost.

Live translation currently spans 20 languages, with 14 new additions including Hindi, Mandarin, Korean, Japanese, and Arabic. Pedestrian navigation—already accessible on Meta's display glasses—is coming to this lineup too. You'll hear "Turn right now" via your frames wandering through a city you don't recognize. That's actually handy.

The Starfire is where this launch gets genuinely hard to read.

Oval frames. Available in black or tortoiseshell. A little gemstone near the camera on the right lens is designed to capture light the way paparazzi flashes do, which is either poetic or ludicrous depending on your tolerance for celebrity mythology. The movable nose bridge is metal instead of plastic, specifically because Jenner stated during design meetings that it was required to wipe off makeup neatly. The charging case contains a mirror inside.

And then there's this: you can set the AI assistant to speak in Kylie Jenner's voice.

That last detail is a trick. A fun one, certainly, but nonetheless a gimmick—the kind that makes for a wonderful demo and gets forgotten by month two. What I think genuinely matters is that Meta is standardizing the idea that your AI helper can have a personality, a voice, a face. Kylie today. Someone else's voice tomorrow. That's the true direction here, even if the current execution is primarily a party trick. At $399, you're paying $100 extra for the jewel, the metal nose bridge, and the bragging rights.

None of this exists in a vacuum.

Worth mentioning first: the recording LED on these glasses—the tiny light that's designed to warn passersby a camera is running—is small enough that security researchers have openly branded it worthless. Modification kits that disable it totally exist and aren't hard to find. That's the baseline trust issue before you even get to the major stuff.

Dormant face recognition code—internally named “NameTag”—has already been spotted sitting inside the Meta AI companion app, which has been downloaded over 50 million times and is required to use any Meta glasses. A June 2026 Wired investigation verified it. Meta scrubbed the code after the story broke, labeled it "merely evidence of exploration," and guaranteed openness before any launch.

That's a cautious way to say, "Yes, we constructed it, and yes, we withdrew it—for now."

Then there's the lawsuit. A federal class action filed in March 2026—Bartone et al. v. Meta Platforms—alleges that human contractors in Kenya were analyzing intimate footage recorded by customers' Ray-Ban Meta spectacles as part of AI training. Bathroom trips, nudity, sexual content—all of it routed through overseas workers who users had no idea existed. Meta's justification was that human review was stated in the terms of service, which is technically correct and practically worthless. Both the UK and Kenyan data regulators launched investigations. Meta has previously paid over $2 billion in biometric privacy settlements, including $1.4 billion to Texas alone in 2024.

At $299, the Adventurer and Fury are the most accessible AI glasses Meta has ever sold and possibly the most accessible in the market right now. The style variety is real—26 combinations is not nothing, and Muse Spark is a legitimately interesting bit of engineering even if you never consciously notice it working. Hindi live translation alone is going to matter to a lot of individuals whom the Ray-Ban line never really spoke to.

The privacy situation is harder. Not impossible to overlook, but tougher. Meta has an established pattern here—design the invasive feature, get caught, delete it, repeat.
The glasses themselves offer completely no way for a bystander to know they're being filmed. That's not a hypothetical concern. It's already been litigated.

If you've been waiting for a cause to try this category, this is perhaps the most reasonable entrance point yet. Just go in knowing what you're buying into, beyond the frame style.