Sent to 23 readers
Race to AGI Daily Digest - Saturday, July 18, 2026
Share:
TLDR
China is using WAICO to push its own global AI governance and open‑source norms, solidifying a separate rules bloc.
The EU’s latest order forces Google to share Android and search data with AI rivals, not just screen space.
US officials are weighing a FINRA‑style AI watchdog, borrowing financial‑market supervision ideas for frontier models.
Moonshot’s 2.8T‑parameter Kimi K3 keeps the open‑weight frontier race alive against Anthropic’s Fable and OpenAI’s stack.
Netflix using generative AI on 300 titles shows multi‑agent workflows are already part of day‑to‑day entertainment production.
The Full Story
Following Monday’s GPT‑5.6 agent stack, Tuesday’s robot funding, and the mid‑week policy wave, today feels like the part where everyone reaches for the rulebook.
China is trying to write its own. WAICO isn’t just a talking shop anymore; Beijing is using it to push new norms for what counts as responsible open‑source and how frontier models should be shared WAICO governance push ->. That locks in Friday’s “global AI blocs harden” storyline: your model’s license now hints at which bloc you’re aligned with.
In Europe, regulators are still leaning on Google. After prying Android open to rival assistants, the EU now wants Android and search data opened up so competitors like Mistral AI -> can train and tune serious alternatives EU data‑access order ->. It’s not just about letting new agents run on your phone; it’s about giving them the behavioral fuel they need.
On the US side, policymakers are workshopping a FINRA‑style AI watchdog to vet frontier models before they hit the wild FINRA‑style AI watchdog story ->. Following Thursday’s reverse‑federalism pitch from OpenAI ->, this is another sign that safety‑by‑training and disclosure are moving from blog posts into possible law.
Meanwhile, the model war keeps humming. Moonshot’s Kimi K3 shows up again as a 2.8‑trillion‑parameter open‑weight frontier model pitched directly against Anthropic’s Fable and other US leaders Kimi K3 frontier model ->, with players like Moonshot AI -> and Anthropic -> now locked into the same comparison set.
And while all that plays out at the top, Netflix quietly says it used generative AI across 300 films and series this year Netflix production stat ->. So the multi‑agent frontier we talked about Monday is already baked into everyday production, even as nurses on Hacker News complain about AI surveillance and Kaggle debates whether “AI slop” should win real prize money.
Here’s the thing: by Saturday, the theme is clear. The big fight isn’t just who has the smartest model—it’s who controls the data, the licenses, and the regulators that decide how those models touch real people.
Get This Delivered Daily
Join thousands of AI professionals who start their day with Race to AGI.