ChinAI NewsletterGreetings from a world where…
a group of puffins can be called a circus, a parliament, or an improbability
…As always, the searchable archive of all past issues is here. Please please subscribe here to support ChinAI under a Guardian/Wikipedia-style tipping model (everyone gets the same content but those who can pay support access for all AND compensation for awesome ChinAI contributors).
Around the Horn (26th episode)
Could we bring back this show in podcast form or something? Let’s go back Around the Horn this week.
This is the 26th roundup of articles on China’s AI ecosystem. For new readers, here’s how it works (see ChinAI #358 for the previous edition):
I give short previews of ten articles that caught my eye during a scan through my usual sources (all published within the past week or so). The title for each preview links to the original article in Chinese.
Readers vote on next week’s feature translation by replying to the email and/or commenting on the post with the number of your preferred article. *Extra weight goes to votes from those of you who support ChinAI through a paid subscription.
The main idea is that any of these 10 links would have made for a great feature translation this week — like the 60 songs that explain the ‘90s podcast, there are no skips!
1) Why AI Geniuses are Flowing Back to Big Tech
Summary: Why have some prominent Chinese young AI talents, including one of the developers of DeepSeek-V2, chosen to go to big tech firms? This article unpacks a variety of explanations, including the collapse of the 2023 boom in large model startups.
Source: 动察 (Beating), outlet that covers finance and technology, under BlockBeats (a blockchain research platform).
2) Most Companion Robots Die by Day 30
Summary: Huxiu convened a roundtable of companion robot startup founders and investors. Some interesting stats: day 30 is a key marker of retention; for some companion robot products, the return rate approaches 30%.
Source: 虎嗅 (Huxiu), a well-known platform that shares user-generated content but also publishes their own pieces on China’s science and technology ecosystem.
3) “In the future, graduating from a ‘985’ or ‘211’ (top tier) university may not matter as much, as careers will depend entirely on one’s actual ability”
Summary: What does the AI era mean for China’s education and workforce training system? Drawn from a lecture and Q&A by Duke Professor Bai Gao, given in Shanghai, this piece explores how AI could challenge education imbalances, with new pathways for students from tier-2 universities and high schools that receive significantly fewer resources.
Source: 知识分子 (The Intellectual) — a platform that covers the state of science in China, founded by Chinese and Chinese-American scientists.
4) Release of AI Safety Benchmark Results for Agents (Claw-class) – Q2 2026 Edition
Summary: The CAICT AI Safety benchmark is worth following closely. This updated evaluation finds that content security is generally controllable (with some hallucination issues), but the agents exhibited high rates of harmful behavior in other areas.
Source: 中国信通院 (CAICT) — a think tank under China’s Ministry of Industry and Information Technology.
5) The First Shot in Domestic Compute for Training: Meituan’s LongCat-2.0 Officially Released
Summary: Outlets have spread misinformation about DeepSeek and Zhipu’s latest models being trained entirely on Chinese chips. That’s just not true. Now, though, Meituan has released the first trillion-parameter model trained fully on a computing cluster of 50,000 Chinese chips. Why did this unlikely player achieve this first?
Source: 雷峰网 (Leiphone) — media portal that covers China’s science and tech landscape, with a focus on AI-related happenings.
6) American AI can no longer hide its ultimate ambition, giving China a major wake-up call
Summary: Ruixiang Li, a researcher at Xiamen University, argues that US AI giants have become quasi-sovereign entities that comprehensively influence public policy. Li contends that China should adopt a different paradigm that upholds national sovereignty and prevent private AI firms from superseding the public interest.
Source: 文化纵横 (Beijing Cultural Review) — leading platform for contemporary political and cultural thought, which also publishes a quarterly journal.
7) After hiding it for three months, Anthropic has removed hidden code that identified “Chinese AI”
Summary: In the April 2026 release of Claude Code, Anthropic quietly added some code aimed at identifying Chinese users. After people figured out what was happening, Anthropic framed the move as an effort to guard against distillation and announced a rollback. In the wake of this revelation, Alibaba has issued an internal mandate to remove all Claude software from employee computers.
Source: China Software Developer Network (CSDN) — the largest software developer community in China, which provides IT news coverage and hosts code
8) Doubao introduces paid tier, ranging from 68 to 5088 RMB: some people left but there are others who want the price to go up
Summary: Responses to the paid tier differed across three groups of users. The first group left: after introducing a paid tier, Doubao’s monthly average users fell by 6 million. The second stayed. Most interestingly, a third group anxiously hoped for higher prices.
Source: 第一新声 (Diyi xinsheng) — Beijing-based information platform focused on the AI value chain, with an affiliated think tank
9) Zhipu vs. MiniMax is China’s projection of the Anthropic vs. OpenAI Rivalry
Summary: When Zhipu and MiniMax debuted on the Hong Kong stock exchange back in January, MiniMax jumped out front with a market cap nearly two times that of Zhipu. Now, Zhipu’s market cap is around six times higher. What are the parallels to the Anthropic vs. OpenAI competition?
Source: Shuzi Lichang (数字力场) — Started following this account back in 2021, but I haven’t browsed them that closely. This article is by Zongming She, a former deputy editor of the commentary section in The Beijing News.
10) The Disappearing Human: is AI taking over the Internet?
Summary: This article grapples with a recent Cloudflare data release: from the sites it hosts, the volume of web traffic requests from machines surpassed that of humans. The author analyzes automated traffic through the context of the Chinese web, which has many “walled garden” super apps that are siloed from the broader internet.
Source: 脑极体 (Naojiti), a tech media platform based in Tianjin. Previous issues have translated their analysis of liquid cooling in data centers.
Thank you for reading and engaging.
These are Jeff Ding’s (sometimes) weekly translations of Chinese-language musings on AI and related topics. Jeff is an Assistant Professor of Political Science at George Washington University.
Check out the archive of all past issues here & please subscribe here to support ChinAI under a Guardian/Wikipedia-style tipping model (everyone gets the same content but those who can pay for a subscription will support access for all).
Any suggestions or feedback? Let me know at chinainewsletter@gmail.com or on Twitter at @jjding99