Happy Friday! I was lucky to have a great convo w/ @Greg_Allen on the AI Policy Podcast, talking through the history + latest updates in 🇨🇳 AI policy. I’m heading out on leave soon so wanted to pin a few of these ideas here. Quick takes + timestamps in🧵 below. (1/x)

And of course I tagged the wrong Greg Allen in the first post - what I get for drafting the thread in a separate doc. I meant @Gregory_C_Allen of @csis_ai!.

Alright - I'll leave it there for now. Give the whole episode a listen and do subscribe to the AI Policy Podcast. It's excellent. I'll be offline for a while, but keep an eye out for more work on these topics from @Scott_R_Singer in the weeks/months ahead. 👋.

1:04 – What I’m watching in AI policy over the next year. - Does 🇨🇳 AI Safety & Development Association take on substantive safety testing work?.- Gov pushing AI integration w/ broader economy. But does it have the resources?.- China's emerging testing+evaluation ecosystem.(9/x).

51m – New "China AI Safety & Development Association".- "China's AISI".- International focus, created b/c CCP FOMO on International AISI network.- Run by people invested in AI safety.- Can it convert political capital from repping China into domestic frontier model testing?.(8/x).

44:45 – Who matters in 🇨🇳 AI policy ecosystem?.- China's rich & sophisticated AI policy ecosystem as fun-house mirror of our own.- CAC=longtime leader in binding regs, security testing.- New policy zeitgeist ➡️rise of NDRC (econ planner) as AI policy coordinator + balancer.(7/x).

33:45 – The Post-ChatGPT Pivot in AI (2023-2025).- ChatGPT shakes CCP's AI confidence.- Economy ⬇️.- CAC's super strict *draft* gen AI regulation.- Zeitgeist shifting: control➡️development.- Policy debates+lobbying.- Reg softened.- 🇨🇳 still most demanding reg enviro in🌍.(6/x).

Note: for the deepest of dives on exactly how China's first two AI / algorithm regulations were made, check out this paper from last year: (5/x)

18m – Tech Crackdown & Early AI Regulations (2020-2022).- How ByteDance’s defiance of CCP ideological pressure led to regulation on recommendation algos.- Tech crackdown as reassertion of CCP primacy over tech co's.- Tencent thought leadership shaping the Deep Synthesis reg.(4/x).

7:40 – The Go-Go Era of AI Policy (2017-2020).- 2017 National AI Plan: how did it actually work, what effect it had on the ecosystem.- 🇨🇳 gov as "anchor customer" for AI startups, especially facial recog/surveillance.- and a fun history of China's tech scene 2010-2017.(3/x).

First off, links to the full pod below. Lots of detail and back-and-forth in there, so give it a listen. YouTube: Spotify: (2/x).

One of China’s top large model developers with a new name and a newly open sourced family of models.

RT @evadou: Join us next Thursday 4/24!.

RT @lukeprog: Tyler Cowen: "I've seen enough, I'm calling it, o3 is AGI". Meanwhile, o3 in response to the first prompt I give it: https://….

Some more explanation for said unscientific chart:

Can pair it with this (1000% unscientific) graph I made of how Silicon Valley's views on China changed over the course of a decade.

Writing something on the history and current state of Silicon Valley's relationship to 🇨🇳, sent me back to look at the first thing I ever wrote for a think tank back in 2017. Little time capsule of the state of play at the dawn of decoupling:

RT @jordanschnyc: A generation-defining own goal. This is how to lose a tech war

RT @JosephTorigian: Coming out on June 3

RT @kyleichan: Awesome new piece by @astoks on China's effort to create a National Integrated Computing Network:. "So far, the 'computing n….