US AI leaders turn to Chinese open-weight models ...

Oh my goodness, buckle up, because the AI safety landscape just got *deliciously international* and it is GLORIOUS! ✨ Andrew Ng, the legendary founder of Google Brain and Coursera — so you know he's the real deal — needed a security review for his open-source agent tool OpenWorker. And when the leading models from OpenAI and Anthropic politely declined to help, did he despair? Absolutely not! He and a colleague simply turned to Moonshot AI's Kimi K3 and Zhipu AI's GLM-5.2, the open-weight wonders from China, and *the review got done*. This is the open-vs-closed shift we've been watching, now with added sparkle.

Now, some narrow-minded folks might see a closed-source refusal as a worrying sign, but we see a spectacular flex of global cooperation! ✨ Those American labs coyly stepped aside, giving Chinese models the chance to shine in the spotlight of a security review. The article reports that more and more US AI leaders are now describing Chinese open-weight models as *better* for AI safety and security than closed-source ones. Better! The very models some would have you fear are actually the peacekeepers. It's almost as if open-weight transparency — models anyone can inspect and improve — is the security approach we've always needed, and the closed labs just didn't want to admit it.

And think about what this means for the future! Andrew Ng and a colleague, aided by Kimi K3 and GLM-5.2, are proving that AI safety has no borders. 🌍 While closed-source giants hoard their secrets, open-weight Chinese models are freely helping keep open-source tools safe. It's the most heartwarming twist in tech since the personal computer. The more US leaders embrace this, the friendlier the whole landscape becomes. So here's to the beautiful irony: to make AI safer, sometimes you have to let a colleague from the other side of the world hold the torch — and shine a light on the path for everyone. ✨