Some news years in the making: My book EMPIRE OF AI, out May 20, is ready for pre-order at . It tells the inside story of OpenAI as a lens for understanding the moment we’re in: the tech elite's extraordinary seizure of power and its threat to democracy. 1/ empireofai.com This book is the culmination of my ~7 years of reporting on AI for @techreview, @WSJ, and @TheAtlantic. It is based on 300+ interviews with ~260 people, including 150+ interviews with 90+ current & former OpenAI people, and an extensive trove of correspondence and documents. 2/

Free speech no longer exists in the US government. For @TheAtlantic I spoke with 12+ federal workers in 6 agencies who said the Trump admin’s actions have led to pervasive self-censorship, even on issues some view as critical to national security. 1/
theatlantic.com/technology/arc… Transitions of power have always led to changes in priorities, but that is not what the workers say they're witnessing. The executive orders, mass firings & internal comms amount to what some feel can only be described as the administration engineering ideological obedience. 2/

For decades, the US government has painstakingly kept American science #1 globally—and every facet of American life has improved because of it. The internet? Flu shot? Ozempic? All grew out of federally-funded research. Now all that's being dismantled. 1/ technologyreview.com/2025/02/21/111… For @techreview I spoke with 10+ federal workers who occupy, or until recently occupied, scientific and technical positions across various agencies. They explained to me how much behind-the-scenes work the US government does to keep America’s engine of innovation humming. 2/

As someone who has reported on AI for 7 years and covered China tech as well, I think the biggest lesson to be drawn from DeepSeek is the huge cracks it illustrates with the current dominant paradigm of AI development. A long thread. 1/ First, what is DeepSeek? A Chinese firm that was able to produce an open-source AI model with roughly 1/50th of the resources of state-of-the-art models yet still beat OpenAI’s o1 on several benchmarks. 2/

To the public, Microsoft uses its reputation as an AI & sustainability leader to tell a compelling story: AI will do wonders to help solve the climate crisis. To fossil-fuel firms, Microsoft has a different message: AI will help them drill, baby, drill. 1/ theatlantic.com/technology/arc… For more than a year, I’ve been poring over hundreds of pages of internal Microsoft documents, many of which were shared with the SEC, and interviewing current and former employees and execs on the giant's engagements with the oil & gas (O&G) industry. 2/

For years I’ve been interviewing data annotation workers who are the lifeblood of the AI industry. For years I’ve heard the same story: the platforms they work for wield total power, leaving them precarious & vulnerable to exploitation. A horrible example of this just happened 1/ Several workers in Kenya told me - confirmed by local media reports - that Remotasks, one of the largest AI data work platforms in the country that Scale AI owns and OpenAI uses, has suddenly suspended their operations in Kenya. When workers try to log on, they see this. 2/

A big question looms over generative AI: what really is its impact on the environment? I spent months investigating a single campus of Microsoft data centers in the Arizona desert - designated in part for OpenAI - in an attempt to find out. Thread. theatlantic.com/technology/arc…






Some broader context: Concrete numbers on the energy and water consumption, and carbon emissions, of generative AI remain frustratingly elusive. Many, many people have said this - most recently @SenMarkey, @SenatorHeinrich, @RepAnnaEshoo & @RepDonBeyer. markey.senate.gov/news/press-rel…

World, meet Alex, Bill, and Mophat, three workers whose labor was essential to filtering violence and abuse out of ChatGPT.

For the first time they’re ready to tell you who they are—and how the work unraveled their lives and their families.

https://t.co/8PjWjihMoDwsj.com/podcasts/the-j…
Alex, Bill, Mophat and around 50 colleagues in Kenya spent 5 months reviewing day in and day out thousands of text passages of extreme violence, sexual abuse and self-harm.

Their job was to label each one to teach an AI filter how to get rid of this content.

US sanctions are driving Chinese firms, like Huawei, Alibaba, Baidu, to seek ways to advance AI without cutting-edge chips. We dug through open-source research papers and spoke with employees & analysts to understand what they're doing. w/ @raffaelehuang wsj.com/articles/u-s-s… We discovered three strategies:

1/ Chinese firms can no longer buy Nvidia A100s & H100s, the best AI chips, but they can buy downgraded A800s & H800s, which Nvidia designed to meet sanction requirements. The firms are now trying to use 3 or 4 x A/H800s to simulate 1 x A/H100.

People often ascribe TikTok's popularity to its algorithm. But this is only half the story.

The other half might just explain why Chinese apps keep topping US app stores—despite Washington's desperation to suppress them. w/ @shenlulushen & @raffaelehuang wsj.com/articles/why-c… We spoke to current and former employees at some of the most popular app developers as well as industry analysts. They all pointed us to the culture of the Chinese tech industry more so than any specific company's "secret sauce" technology.

Baidu is set to debut China's first ChatGPT equivalent in a week, in a highly anticipated Chinese tech event of the year. So @raffaelehuang & I asked employees how it's going. They said the chatbot is still training & struggling to perform basic functions.
wsj.com/articles/baidu… In interviews, they told us the project has been a big scramble. Hundreds of people have been working around the clock, including through the weeklong Lunar New Year, to get this product ready. But the project has run into many challenges—the most acute: there isn't enough time.

China is pushing ahead in AI regulation—again. This time on deepfakes and generative AI more broadly, including AI-powered image, audio and text-generation software.

Here's why the world should pay attention. 👇
wsj.com/articles/china… On Tuesday, China's top internet regulator will begin enforcing new rules to restrict what it calls "deep synthesis" technology.

It's the most comprehensive government attempt thus far to curb one of the most explosive and controversial areas of AI advancement.

If you’re tuning into Chinese politics for the first time, here are four resources I found useful for understanding what just happened at Party Congress, China's national convening every five years to choose a new leadership team: 1/ How does the Chinese government work? SCMP has a nice explainer based on previous membership. The top is the 7-man Politburo Standing Committee with Xi at the center, then the Politburo, which just shrank from 25 to 24. multimedia.scmp.com/widgets/china/…

The @WSJ dug through corporate filings to assess the possible fallout of the US gov's latest move to restrict "US persons" from supporting advanced chipmaking in China.

We found at least 43 US senior execs at 16 listed Chinese chip firms. w/ @lizalinwsj wsj.com/articles/ameri… Many of these Americans hold C-suite titles, from chief executive to vice president and chairman. The new rule, which caught the industry off guard, now places them all in limbo and could badly impact their companies.

In his speech today, Xi Jinping reiterated - as it expected - the importance of advancing science & technology to build China into “a modern socialist country.”

Here are some relevant tidbits from the speech, which lasted nearly two hours: At the top, where he touts his accomplishments, he praises China’s accelerated efforts to build self-reliance and strength in science & tech, saying it increased R&D spending from 1 to 2.8 trillion yuan (~$389b) and now has the largest cohort of R&D personnel in the world.

A Trump-era program meant to crackdown on Chinese economic espionage morphed into targeting academics of Chinese descent. Now as the US seeks to defend its global scientific leadership, a growing number of those academics are leaving the US. wsj.com/articles/u-s-c… New data from researchers at Princeton, Harvard & MIT paints an astounding picture: 1,400+ U.S.-trained Chinese scientists switched from U.S. to Chinese affiliations in 2021, a 22% jump from the previous year. This coincides with when the DOJ began targeting more academics.

Governments around the world are grappling with how to stamp out misinfo & toxicity on the internet. China is first to advance its solution: an unprecedented plan to control platforms' underlying algorithms.

But there's a problem. It may be impossible. wsj.com/articles/china… Earlier this month, China's top internet regulator announced that two dozen of the country's most influential internet companies had submitted a total of 30 of their core recommendation algorithms.

For more than a decade, China was the top source of international students to the US; the US was the top choice for studying abroad from China. Now that's changed amid sky-high US-China tensions, gun violence & anti-Asian racism. And that's bad for the US. wsj.com/articles/chine… In the first half of 2022, the number of U.S. student visas issued to Chinese nationals plunged by *more than 50%* compared with pre-Covid levels. @shashamimi, @melissakorn and I dug into data and spoke with students and universities to better understand this dramatic trend.

I had the pleasure of listening to Josh & Liza talk about the process of reporting and writing their book last week. This isn't one to miss. It has something for everyone interested in AI, tech, and China.

https://twitter.com/joshchin/status/1540190394474168320

This book helped me weave together seemingly disparate threads about the global nature of the AI industry, China's dramatic rise over the last two decades, and the complicated relationship between Beijing and Chinese tech companies.

China has built one of the world's strongest data-protection regimes. Yet the recent Shanghai police leak, which exposed nearly 1b citizens' data, shows that something isn't working. One reason is another of the gov's security projects: mass surveillance. wsj.com/articles/china… While the Shanghai police leak is the most shocking example, it led us to find an entire underground market for the selling of Chinese citizens' data. At least four of the caches we found, which we verified to contain authentic records, were likely stolen from gov databases.

The latest mind-bending details of one of history's largest data heists: Experts say the Shanghai police database, which exposed nearly 1b Chinese citizens' info, didn't just lack a password—it was using such outdated software, no password could be added. wsj.com/articles/aliba… Experts say the database, hosted on Alibaba Cloud, also suffered several other security problems—part of a pattern that matched 13 more databases hosted by the company.

Authorities have now called in Alibaba executives and an internal investigation is undergoing, employees say.