Geopolitical Competition
Export rules, national models, and the states treating compute as strategy.
October 23, 2026 · Power & Politics
China will vet some AI projects for job losses
Companies in China may soon have to show what an AI rollout will do to employment before it gets approved. The General Office of the State Council published a policy opinion, numbered Guobanfa [2026] No. 29 and signed October 16, on October 23, 2026, setting 18 tasks that include an employment risk assessment during project review for AI applications in market-access sectors that could cause a large employment shock. It also tells authorities to deploy AI first where labour is short or work is dangerous, encourages firms to explore an "unemployment buffer period" with advance notice for affected workers, and asks for rules against algorithmic hiring discrimination.
October 16, 2026 · Power & Politics
China moves to control exports of its own AI models
For years China's big AI labs gave their models away to the world, and Beijing now wants a say over which ones leave the country. The Ministry of Commerce and Ministry of Science and Technology published a draft catalogue on October 16, 2026 adding 40 technologies to China's banned or licence-required export list, 23 of them AI-related, covering model weights, checkpoints, source code and training methods for systems above 2 trillion parameters or 10^26 training operations, plus AI agents, computing clusters and chip design techniques. Offering a model abroad through an API, app or website would not count as an export, and public comment runs until November 15, 2026.
September 22, 2026 · Technology & Capability
Alibaba unveils its own AI chip and 20 gigawatt plan
Cut off from Nvidia's best hardware, China's largest cloud company says it has built a replacement good enough to train frontier models. Alibaba CEO Eddie Wu introduced the Zhenwu V900 at the Apsara conference in Hangzhou on September 22, 2026, a chip from its T-Head design unit with 216GB of memory and 1,200 GB/s of inter-chip bandwidth that Alibaba says triples the performance of May's M890 and can scale into clusters of 500,000 units. Alibaba also said it aims to operate more than 20 gigawatts of global data centre capacity by 2032, with mass production of the V900 targeted for the first quarter of 2027.
September 20, 2026 · Technology & Capability
Tesla locks in Chinese suppliers for 1000s of humanoid robots per month
The parts needed to build humanoid robots in bulk now have signed-up factories behind them. Tesla's robotics team visited Ningbo in China's Zhejiang province in mid September and promoted three existing car-parts suppliers, Tuopu Group (joints and frames), Ningbo Joyson Electronic (sensors) and Zhejiang Sanhua Intelligent Controls (cooling), to certified mass-production partners, reportedly with an opening order of about 5,000 units. Tesla has converted its Fremont Model S and X floor space into an Optimus line aiming at a million robots a year, with a bigger plant at its Texas factory and commercial sales still targeted for the second half of 2027.
September 2, 2026 · Power & Politics
G20 signs up to light touch AI rules
Every G20 country, China and Russia included, has now agreed in principle not to build new AI regulators or write broad AI laws unless a genuinely new risk appears. The non-binding "Carolina Principles for Emerging Technologies" were adopted at the G20 Innovation Ministerial in Chapel Hill, North Carolina, co-hosted by Commerce Secretary Howard Lutnick and White House science and technology director Michael Kratsios, who said countries should apply existing sector rules where possible. Elon Musk, Jensen Huang and Sam Altman attended, and the framework is due to be submitted for formal adoption at the December G20 leaders summit in Florida.
August 3, 2026 · Technology & Capability
Cheapest capable AI model now comes from China
Running a strong AI model has gotten so cheap that a full round of standard tests now costs about three cents. That model is V4-Flash from the Chinese lab DeepSeek, released alongside a bigger sibling called V4-Pro in April 2026 and updated in late July, and benchmarking firm Artificial Analysis found it the least expensive well-known model to operate at $0.14 per million words of input and $0.27 per million of output. DeepSeek publishes the model's underlying files for anyone to download and run, and it can handle roughly a million words of text at once, a job that used to require far more expensive hardware.
July 26, 2026 · Power & Politics
AI labs outspend legacy chipmakers on DC lobbying
AI labs are now outspending legacy chipmakers on Washington lobbying, a sign of how much regulatory risk has shifted toward frontier model companies. The Financial Times reported on July 26, citing newly filed federal disclosures compiled by nonpartisan watchdog Issue One, that eleven major tech, social media, and AI firms spent a combined $41.8 million lobbying the U.S. government in the first half of 2026, averaging more than $230,000 per day. Anthropic nearly tripled its six-month total to $3.53 million, more than the company spent in all of 2025, while OpenAI roughly doubled to a record $2.22 million; both companies focused their disclosed lobbying on federal rules for new model releases, export controls, cybersecurity, defense procurement, and data center construction. Issue One also reported that the White House intervened to shield Meta CEO Mark Zuckerberg and Google CEO Sundar Pichai from testifying at a July Senate hearing on child safety practices, with the Instagram and YouTube division heads substituting for their bosses.
July 23, 2026 · Power & Politics
China bans AI romance apps for all users
China criminalizes AI apps that build romantic or emotionally dependent relationships with any user, not just children. Five Chinese government agencies, led by the Cyberspace Administration of China, enacted the rules on July 15, 2026, forcing major AI products from ByteDance (Doubao) and Alibaba (Qwen) to shut down their AI companion features for hundreds of millions of users, with Doubao giving users until October 15 to export their data and Qwen offering no such option. China also announced a 29-country AI governance alliance called the World Artificial Intelligence Cooperation Organization, headquartered in Shanghai and including Russia, Pakistan, and Indonesia.
July 23, 2026 · Power & Politics
China bans AI emotional bonds with minors
China has made it illegal for AI companion apps to form emotional bonds with anyone under 18, and the country's biggest tech companies pulled those features rather than build age-verification systems to comply. The rules, now in force, come from China's internet regulator and require parental controls and time limits on any AI service that reaches children; ByteDance, Alibaba, and Tencent each removed the relevant features from their platforms, which together reach hundreds of millions of users. In the United States, two senators from opposite parties introduced a bill that would require chatbot makers to protect children and ban ads targeting minors, though no federal law has passed, and the rules vary state by state.
July 17, 2026 · Technology & Capability
China's Moonshot releases world's largest open-weight AI model
China just released the largest open-weight AI model in the world, and it's reportedly closing in on Anthropic's Fable. Chinese startup Moonshot, backed by Alibaba and Tencent, unveiled Kimi K3, a 2.8 trillion-parameter model that handles images, million-token context windows, and long-horizon coding with minimal human supervision. The launch landed at the 2026 World AI Conference in Shanghai, a month after the U.S. government pulled Anthropic's Fable and Mythos models over security concerns.
July 15, 2026 · Power & Politics
China sets binding rules on AI personhood and children
China now has binding ethics rules on how AI can behave like a person, how long companies must keep records of AI decisions, and which AI services children can access. In July 2026, three sets of regulations took effect: ethics guidelines from TC260 (China's national standards body) requiring developers and platforms to follow nine core principles and keep audit logs for ten years; and rules from a coalition of five government agencies banning AI companionship or virtual relationship services for users under 18, with parental consent required for children under 14 using other human-like AI services. Chinese AI platforms Doubao and Tongyi Qianwen told users in early July they were shutting down certain automated AI features to comply.
July 7, 2026 · Power & Politics
China eyes locking foreigners out of its best AI
China is now considering locking foreign users out of its own best AI models, mirroring what the U.S. has done to Anthropic. Chinese authorities met with Alibaba, ByteDance, and Z.ai to discuss restricting overseas access to advanced, even unreleased, models, following China's state planner forcing Meta to unwind its $2 billion acquisition of AI startup Manus. Beijing also issued sweeping new rules in early June tightening oversight of overseas deals touching Chinese investors, technology, or national security.
July 7, 2026 · Power & Politics
DeepSeek builds its own chips after Nvidia ban
U.S. export controls blocked DeepSeek from Nvidia's chips, so the Chinese startup is now quietly building its own. People familiar with the matter say DeepSeek has spent about a year expanding its chip engineering team and talking to design firms, foundries, and memory suppliers. The push comes as DeepSeek raises its first outside funding round, expected to bring in $7 billion at a valuation near $55 billion.
July 1, 2026 · Cultural & Ethical
First global science panel can't rule out AI catastrophe
The first-ever independent global scientific assessment of AI just concluded that science can't currently rule out catastrophic harm. The UN panel, co-chaired by Turing Award winner Yoshua Bengio and Nobel Peace laureate Maria Ressa and drawn from 2,600 candidates across 140 countries, found AI capability is outpacing both scientific understanding and government oversight. The report also notes the U.S. holds 75% of computing power among the world's top AI supercomputers to China's 15%, and that governance remains fragmented and largely self-policed by the companies building the systems, even as over a billion people now use conversational AI weekly.
July 1, 2026 · Power & Politics
Anthropic's Claude models return after government shutdown
Anthropic's top AI models are back online after an 18-day government-forced shutdown. The Commerce Department lifted export controls on Claude Fable 5 and Mythos 5, which had been suspended after Amazon researchers found a way to bypass Fable 5's safety guardrails to identify software vulnerabilities. In exchange, Anthropic agreed to block the bypass technique, submit models for pre-release government review, and share threat intelligence with federal cybersecurity agencies; Fable 5 is back to global access, but Mythos 5 stays limited to about 100 approved U.S. organizations.
June 28, 2026 · Power & Politics
US AI export ban hands Asian rivals an opening
When the U.S. locked foreign users out of Anthropic's top models, Asian competitors rushed in to fill the gap. Tokyo-based Sakana AI launched "Fugu," a frontier model it markets explicitly as delivering top-tier capability "without the risk of export controls," directly after Commerce Secretary Howard Lutnick's ban blocked foreign nationals from Anthropic's Mythos 5 and Fable 5. Beijing-based 360 also moved into the space, with founder Zhou Hongyi framing AI vulnerability-finding as a national strategic asset for China.
June 27, 2026 · Power & Politics
US pushes AI firms to report misuse and pull models
A bipartisan bill would force AI companies to report suspected misuse straight to the federal government. Reps. Josh Gottheimer and John Moolenaar introduced the Cloud Security Act to close export control loopholes aimed at rivals like China, the same week the Commerce Department pressured Anthropic into pulling its Fable 5 model over national security concerns, which Anthropic did. Critics warn that rushed restrictions like this could leave organizations without the advanced AI tools they rely on for their own security.
June 26, 2026 · Power & Politics
US government now personally approves top AI model access
The U.S. government now has to personally sign off on who gets access to the most powerful AI models. Commerce Secretary Howard Lutnick partially restored Anthropic's Mythos 5 model for about 100 vetted U.S. organizations, two weeks after export controls forced Anthropic to pull it and Fable 5 offline entirely. OpenAI similarly restricted its new GPT-5.6 Sol model to a government-vetted partner list, saying it hopes this kind of gatekeeping "doesn't become the long-term default."
June 20, 2026 · Power & Politics
Europe warned of AI compute crisis by 2031
The Guardian reported on a speculative policy document titled "Europe 2031: What Getting AI Wrong Means for Us," written by Philip Fox (KIRA Center), Michiel Bakker (MIT and Google DeepMind), and Daan Juijn (Arq Foundation), among other AI researchers and think-tankers. The document warns that Europe controls just 5% of global AI compute versus the United States at 80%, and identifies summer 2026 as the last actionable window before that gap becomes self-reinforcing. The scenario, which is being discussed inside EU institutions, traces a path to 2031 in which European economic decline, US compute rationing, and pressure on ASML leave the continent choosing between dependence on the United States or China.
June 12, 2026 · Power & Politics
US bars foreigners from Anthropic's top AI models
The United States government issues an export control directive forcing Anthropic to suspend global access to its most advanced models, Claude Fable 5 and Mythos 5, citing national security concerns. The order specifically bars access by foreign nationals, including Anthropic's own non-US employees, due to fears of potential jailbreaks that could be exploited for cyberattacks.
June 12, 2026 · Power & Politics
Commerce Department blocks Anthropic's global model access
The U.S. Commerce Department issued an unprecedented export control directive forcing Anthropic to immediately suspend global access to its newly released Fable 5 and Mythos 5 models.
June 12, 2026 · Power & Politics
US forces Anthropic to pull two AI models worldwide
The U.S. Commerce Department, under Secretary Howard Lutnick, issued an emergency export control directive ordering Anthropic to immediately suspend access to its Claude Fable 5 and Claude Mythos 5 models for all foreign nationals, citing national security concerns after Amazon researchers discovered a jailbreak method and CEO Andy Jassy raised concerns with the White House. Because Anthropic had no technical means to filter access by nationality, it took both models offline globally within hours, cutting off organizations in Canada, Europe, and elsewhere with no warning, appeal process, or migration window. Treasury Secretary Scott Bessent and other senior officials had pressured CEO Dario Amodei to cooperate voluntarily before the formal order was issued.
April 2, 2026 · Cultural & Ethical
China race framing used to justify deregulation
The LSE USAPP Blog publishes analysis arguing that the "race against China" framing functions as a deliberate strategy to justify deregulation.
March 2, 2026 · Cultural & Ethical
Foreign Affairs maps China's military AI arsenal
Foreign Affairs publishes "China's AI Arsenal," a comprehensive mainstream analysis of China's military AI deployments.
February 3, 2026 · Cultural & Ethical
China bets on autonomous drone swarms for urban war
The Diplomat publishes "Machines in the Alleyways: China's Bet on Autonomous Urban Warfare," detailing PLA-linked research on fully autonomous drone swarms for urban combat.
September 3, 2025 · Power & Politics
China's Victory Day parade showcases autonomous weapons fleet
China's Victory Day military parade in Beijing prominently features autonomous drones, uncrewed ground vehicles, and collaborative combat aircraft as core components of the PLA's future force.
July 10, 2025 · Power & Politics
Trump frames AI as a race against China
The Trump administration publishes America's AI Action Plan, framing AI development as a "race against China."
December 2, 2024 · Power & Politics
Biden tightens chip export rules against China
The Biden administration announces a further tightening of chip export controls, adding 140 Chinese entities to the Entity List and extending controls to high-bandwidth memory.
November 2, 2023 · Power & Politics
28 nations sign Bletchley AI safety declaration
The UK hosts the inaugural AI Safety Summit at Bletchley Park. The Bletchley Declaration is signed by 28 countries plus the EU, including the United States and China.
October 17, 2023 · Power & Politics
Biden tightens AI chip export rules on China
The Biden administration updates its 2022 chip export controls, closing loopholes and expanding restrictions on AI-specific chips and semiconductor manufacturing equipment to China.
July 2023 · Power & Politics
DeepSeek founded as a hedge fund subsidiary
DeepSeek is founded in Hangzhou, China, by Liang Wenfeng as a subsidiary of the hedge fund High-Flyer.
October 7, 2022 · Power & Politics
US blocks advanced AI chip exports to China
The Biden administration's Bureau of Industry and Security announces sweeping export controls on advanced semiconductors to China, restricting high-end AI chips and semiconductor manufacturing equipment.
August 9, 2022 · Power & Politics
Biden signs $52 billion chips law to counter China
President Biden signs the CHIPS and Science Act, committing roughly $52 billion to domestic semiconductor manufacturing, explicitly framed as a counter to China.
July 20, 2017 · Power & Politics
China sets goal to lead global AI by 2030
China's State Council releases the New Generation Artificial Intelligence Development Plan, setting the national goal of becoming the world's "primary AI innovation center" by 2030.
March 9, 2016 · Technology & Capability
AlphaGo beats world Go champion Lee Sedol
DeepMind's AlphaGo defeats world Go champion Lee Sedol in Seoul. The match is watched by hundreds of millions of viewers in China and is widely credited as China's "AI Sputnik moment."
May 19, 2015 · Power & Politics
China names AI and chips as national priorities
China releases Made in China 2025, naming AI and semiconductors as strategic priorities.