Meta's next AI glasses drop the camera and keep six mics
Two answers landed on the same afternoon in San Francisco: what Meta wants on your face next, and what the chip industry's loudest voice thinks about the rules for building it.
Meta plans to release a camera-free pair of smart glasses this fall, internally code-named Luna, and is targeting its Connect developer conference on September 23–24 to show them, with shipments in October, The Information reported Tuesday. Reuters, which corroborated the report, lists the specifics: six built-in microphones so the wearer can talk to Meta's AI chatbot and to Muse, the company's consumer agent; a side button that summons Meta AI; and speakers carried over from the current models. Meta did not respond to a request for comment, so treat the product as reported rather than confirmed — though the timeline is unusually short for hardware, and Meta has already booked the stage.
Why it matters: the camera has been the entire fight, and Meta is answering a different question than the one being asked. After a year of "pervert glasses" reputational damage, cinema chains restricting the devices over piracy fears, and Norway weighing a ban on camera-enabled wearables, Luna removes the lens and keeps the listener. Six microphones in a frame is an always-available input device with no shutter, no recording LED, and no obvious tell that anything is being captured as audio — and the mics exist to feed Muse, the consumer agent Meta has been pushing onto phones since August, which we covered when Meta shipped Muse inside a sealed VM. Dropping the camera is a real concession on the loudest objection, and a quiet expansion of the surface Meta can listen from. If the model ships as reported, watch whether the privacy argument moves to audio — and whether regulation that was written about cameras can reach a device that never takes a picture.
Nvidia's Jensen Huang used the same conference to tell the labs they do not need new laws to slow the frontier down — a direct rejection of Anthropic's request for antitrust cover to coordinate on safety. Asked on CNBC about proposals to exempt AI companies from antitrust restrictions so they could jointly pace development, Huang called the idea "completely unnecessary": "We have plenty of laws. We have plenty of regulations that govern the reliability and the functionality of products." His alternative was engineering — "safety is an engineering problem, testing is an engineering problem" — with labs simply holding a product back until it passes, and a flat dismissal of the extinction framing: "We're not going to die in 2030." The BBC reported the same stance, with Huang adding that industry leaders, not legislators, are the ones who should decide when a model ships.
The structural point is who is talking. Nvidia sells the compute that every one of these labs burns, so a slowdown costs Huang's customers money before it costs him anything — and he still refused to endorse the legal mechanism the labs say they need. That leaves Amodei's proposal, published Saturday, in a narrower corner than when it launched: the third-party evaluators he wants embedded at each lab are the easy half, while the coordination that requires government support is the half Washington has now answered twice in a day, with Treasury refusing a liability shield and the FTC chair calling the safety-for-antitrust trade a request for barriers to entry — the FTC's answer is here. Amodei, who appeared on stage at Dreamforce the same day, said Anthropic is now in "a dialogue with the rest of the industry" about standards. A dialogue is not a waiver.
What to watch: whether any of the three labs turns the standards dialogue into a published document before Meta's October shipments, and whether Luna's launch materials mention audio retention at all.
If the objection was the camera and not the microphone, does removing the camera settle anything? Tell us in the comments.
