Meta explores neural control and AI defeating bot detectors

Welcome to ZDNET Innovation indexidentifies the most innovative developments in technology from the past week and ranks the top 4, based on votes from our editorial board and experts. Our mission is to help you identify the trends that will have the biggest impact on the future.

Meta leads this week with released Orion, its new AR glasses. Launched as a prototype at Meta ConnectionThey initially impressed ZDNET editor Kerry Wan for being more clear AR experience than Professional vision so far there have been. Instead of “capturing and recreating what’s in front of you,” as Wan puts it, Orion uses holograms to visualize incoming messages and other notifications, keeping the wearer socially aware instead of being stuck in their headphones. What stands out most, however, is the promise of an included neural interface that can interpret finger gesture commands.

Also: In an unexpected twist, Meta suddenly crushed Apple in the war of innovation

Meanwhile, at number 2, Swiss researchers successfully trained an AI model to complete reCAPTCHA tests — you know, those picture quizzes meant to tell humans from robots — with 100% accuracy. Points, bots. While no one is too concerned right now, this development makes reCAPTCHA seem a bit outdated as a browser security measure. Verification checks will become increasingly more difficult, or discreet on-device behavior monitoring will become more important in preventing malicious activity. Neither option feels good for user experience or data privacy in the long run.

Third again is Meta — the same company upgrading its existing Ray-Bans with the ability to “remember” what you look at and save information for later use. The glasses are intended to give a smooth, natural feel artificial intelligence including the increasingly popular live translation capabilities, accessibility perks for those with visual impairments, and the ability to remember where you parked your car (so you don’t have to). These upgrades make everyday AI wearables more common — although that seamlessness also means that specs are always being watched and listened to.

Rounding out the week is OpenAI’s Sam Altman, who published a breathtaking article about “superintelligence” only “a few thousand days left”, loosely implied artificial general intelligence (AGI). ZDNET contributor Tiernan Ray was quick to address the incident, citing some academic concerns to the contrary, along with some critics calling the comments manipulative.

Also: I held the world’s thinnest folding phone in my hand and it made my iPhone 16 Pro Max feel outdated

But why all the fuss about Altman’s optimism, you ask? His comments come at a crucial time for the AI hype cycle; some, like ZDNET writer Taylor Clemons, think the “AI bubble is about to burst”. By popularizing an endlessly positive view of AI’s ability to heal the world (forget all the question marks around social and environmental impact, bias, and scalability), Altman has the risk of pushing that doubt too close to the edge of the abyss.