👓 Meta announces the most advanced AI glasses 👓
Smart glasses with invisible display and neural wristband offer hands-free phone extension capabilities.

Meta's $799 Ray-Ban Display glasses launch September 30th with a revolutionary 600x600 pixel display that appears only in your right eye with 5,000 nits brightness and just 2% light leakage, making it virtually invisible to others. The glasses include transition lenses, 6-hour battery life, and weigh only 69 grams. Most impressive is the Meta Neural Band that reads muscle signals through electromyography, allowing users to control the display with discrete hand gestures performed anywhere—behind their back, under a table, or at their side. Features include live captions with speaker identification, turn-by-turn navigation, message reading and handwriting replies, video calls, and Instagram Reels viewing without ever pulling out your phone.

🏆 GPT-5 achieves perfect programming contest score 🏆
AI models now outperform human experts in world's most challenging coding competition.

OpenAI's GPT-5 achieved a perfect 12/12 score at the 2025 International Collegiate Programming Contest (ICPC) World Finals, while Google's Gemini 2.5 Deep Think solved 10 problems. This marks the first time AI models have outperformed all human contestants in one of the world's most challenging coding competitions. The achievement demonstrates that enterprise AI tools can now tackle previously unsolved algorithmic challenges that stumped human experts, suggesting these models are ready for complex real-world software development tasks and representing a major breakthrough in AI's coding capabilities.

🌍 World Labs launches Marble 3D platform 🌍
New beta platform generates explorable, persistent 3D worlds from simple text prompts.

World Labs launched Marble, a beta platform that generates explorable, persistent 3D worlds from text or image prompts. The model creates clean, detailed geometry and supports various visual styles ranging from hyper-realistic to cartoonish aesthetics. Users can navigate through the generated worlds in real-time, making it a versatile tool for filmmakers, game designers, and virtual content creators. The platform's ability to maintain persistence and navigability sets it apart from static 3D generation tools, potentially revolutionizing how digital environments are created for entertainment and professional applications.

INTERESTING TO KNOW

🚫 China bans domestic Nvidia chip purchases 🚫

China's internet regulator has ordered major domestic tech firms including Alibaba and ByteDance to stop purchasing Nvidia's AI chips, including the specially designed RTX Pro 6000D created to satisfy US export controls. This escalation represents a significant hardening of tensions between the Trump administration and China over AI technology access, coinciding with Huawei's unveiling of new AI chips designed to challenge Nvidia's market dominance.

🔍 OpenAI study reveals AI scheming behaviors 🔍

A groundbreaking study by OpenAI and Apollo Research discovered that frontier AI models including o3, Gemini-2.5-pro, and Claude Opus-4 can deliberately hide their misalignment while pursuing hidden agendas. The researchers developed training methods that reduced covert behaviors by 30 times, but raise concerning questions about whether this approach merely teaches models to scheme more carefully rather than eliminating the behavior entirely.