Full Summary
This Friday morning, major AI developments are unfolding, with both Chinese tech giants and OpenAI making significant moves. ByteDance is reportedly training an AI model with an astonishing 10 trillion parameters, a figure confirmed by both The Standard (HK) and NewsBytes. This model would be more than three times the size of Moonshot AI's Kimi K3, and nearing the scale of Anthropic's Mythos 5, estimated at 8 trillion parameters. Meanwhile, OpenAI is delaying the release of its new Astra model due to internal concerns about its advanced cybersecurity capabilities. Benzinga and The New Stack both report that internal tests suggest Astra may reach a "Critical" threat level, meaning it could create zero-day exploits or carry out complex cyberattacks. OpenAI states it cannot rule out these critical capabilities, prompting stricter security measures and expanded testing in isolated environments. Alibaba is also in the spotlight, launching its new Qwen3.8-Max AI, which boasts 2.4 trillion parameters and tops Chinese text-model rankings, according to Crypto Briefing. Both marketscreener.com and MIT Sloan Management Review Middle East confirm Alibaba's plan to introduce a revenue-sharing model for large commercial users of this open-weight model, signaling a new monetization strategy for advanced AI. In other news, a 4-billion-parameter model from Neon and Castform reportedly matched GPT-5.6 Sol on a specific task at 100 times lower cost, highlighting the rise of specialized, efficient AI. On a concerning note, Tech Times reports that Stanford University's AI, Evo 2, has designed functional, replicating viruses, exposing a significant biosecurity gap. Finally, in the realm of practical applications, both the US and China are launching new AI models for weather prediction, aiming for faster, cheaper forecasts. The US has Project EAGLE, while China introduces the Fengyuan V1.0 model. This means your next weather forecast could be more accurate and timely, but the rapid, sometimes alarming, advancements in AI across sectors demand heightened awareness about both its potential and its risks.