Meta’s highly anticipated launch of its new AI-powered smart glasses at the 2025 Meta Connect event was meant to showcase cutting-edge technology redefining human-computer interaction. However, the live demos took an unexpectedly rough turn marked by multiple technical failures, leaving the...
Meta’s highly anticipated launch of its new AI-powered smart glasses at the 2025 Meta Connect event was meant to showcase cutting-edge technology redefining human-computer interaction. However, the live demos took an unexpectedly rough turn marked by multiple technical failures, leaving the audience and Meta’s CEO, Mark Zuckerberg, visibly struggling to maintain momentum. Here is a detailed rundown of what unfolded during this high-profile unveiling.
Key Highlights:
-
Meta’s AI glasses failed twice in major live demo glitches during the Connect 2025 keynote.
-
A cooking demo using the glasses’ Live AI assistant faltered when the AI jumped steps erratically.
-
Meta’s internal network was inadvertently overwhelmed, causing cascading Live AI failures.
-
Zuckerberg struggled to demonstrate WhatsApp video calls on the glasses.
-
Meta’s CTO confirmed the glitches originated from a “self-inflicted DDoS” on their development server.
-
Despite technical setbacks, Meta’s candid live presentation was noted as refreshingly honest.
The Demo’s Rocky Start
The event opened with enthusiasm as Zuckerberg introduced the second generation of Meta Ray-Ban smart glasses, with promises of seamless AI assistance, longer battery life, and smarter interactive features. The crowd anticipated a smooth demonstration of the AI’s ability to interpret surroundings and provide real-time assistance.
However, the first hiccup arose during a cooking segment where food influencer Jack Mancuso attempted to follow a Korean-inspired steak sauce recipe with help from the Live AI assistant embedded in the glasses. Instead of fluid guidance, the AI repeatedly gave incorrect or skipped instructions, causing confusion on stage. Mancuso hinted at Wi-Fi issues, while Zuckerberg lightly joked about tech demo unpredictability, though the real problem was deeper.
Behind the Scenes: The Cause of the Failures
According to Andrew Bosworth, Meta’s Chief Technology Officer, the demo issues were not due to simple connectivity problems but stemmed from the system architecture itself. When Mancuso activated the Live AI command, it unintentionally triggered every AI-enabled pair of Meta glasses within the venue. Because Meta had routed all Live AI traffic to a dedicated development server for isolation, this caused a denial-of-service (DDoS) type overload on the server, incapacitating the AI globally for the demo .
Zuckerberg’s Call Demo Glitch
A second major flub occurred when Zuckerberg tried to showcase the glasses’ ability to take video calls via WhatsApp. Despite multiple attempts, the glasses’ neural interface failed to register the incoming call. Bosworth explained this was due to a rare bug triggered when the glasses went into sleep mode just as the call alert arrived. This elusive issue has since been fixed but embarrassed the live event’s flow.
Meta’s Transparency Amidst the Chaos
While the demo failures were dramatic, skipping the usual polished pre-recorded showreels underscored Meta’s commitment to transparency and real-world testing. Zuckerberg’s candid responses and technical explanations provided a refreshing honesty rarely seen in major tech unveilings, contrasting with other giants’ guarded presentations.
Looking Ahead
Despite the rough live debut, the Meta AI glasses remain a potentially transformative product, marrying AI with wearable augmented reality. Meta continues to invest heavily in refining the technology and infrastructure supporting these devices. The glitches served as a stark reminder of the challenges in pioneering futuristic tech and the unpredictable nature of live demos.
Source: The Verge, Hindustan Times, Road to VR, and Engadget.
Show more
