China resists calls to slow AI
Why China is pushing back on US warnings over rapid AI development
Monday, 21 September 2026
Warm-up
- What worries you most about rapid AI progress?
- When should governments limit new technology, if ever?
- How do export controls affect innovation and rivalry?
Vocabulary
- receptive
- Willing to listen to or accept a new idea or suggestion.
- sovereignty
- A state's independent power to control its own affairs.
- export control
- A rule that limits sales of certain goods to other countries.
- fearmongering
- Spreading fear to influence opinions or actions.
- governance
- The way rules and decisions are made and enforced.
- iteration
- One version in a series of repeated updates or designs.
- recursive
- Able to act on and improve itself again and again.
- autocracy
- A system where power is held by one ruler or a small group.
Reading
China is pushing back against US warnings about rapid AI development. In Silicon Valley, insiders have warned of a "loss of control" over frontier systems. Dario Amodei of Anthropic proposed a worldwide pacing of the frontier that would require cooperation with China, and continued export controls on advanced chips. He argued democracies should keep their lead over autocracies.
Analysts say Beijing is not receptive. Lian Jye Su at Omdia argues the government treats AI as a core sovereignty capability and believes slowdown calls aim to lock in a US lead. China's foreign ministry said fearmongering, confrontation and vicious competition would disrupt global AI governance.
Not only has Beijing tightened oversight, it has also released a third iteration of its AI safety governance framework. Gabriel Wagner of Concordia says it maps major risks, including agents, cybersecurity threats from frontier models, and the recursive self-improvement of future systems. He sees more common ground than many assume.
By contrast, US president Donald Trump called wipeout fears a hoax. He is due to meet Xi Jinping in Washington next week, though few expect a breakthrough.
Comprehension
- What global proposal did Dario Amodei make?
- Why does Beijing reject a slowdown, according to Lian Jye Su?
- What new AI safety document did China release?
- Which three risks does Gabriel Wagner highlight?
- Where are Donald Trump and Xi Jinping expected to meet?
Grammar focus
Adding emphasis - inversion with negative adverbials
Start a sentence with a negative adverbial (Never, Rarely, Seldom, Under no circumstances). This causes inversion: auxiliary + subject + main verb (e.g. 'Never have they ...'). (Examples constructed.)
- Never have analysts warned of faster AI pacing.
- Rarely has a country tightened oversight so quickly.
Grammar exercise
Circle the correct option in each sentence.
- Never (have observers seen / observers have seen) such rapid change.
- Rarely (a government has tightened / has a government tightened) oversight.
- Seldom (do analysts agree / analysts do agree) on policy paths.
- Under no circumstances (companies should / should companies) ignore risks.
- Only then (did leaders act / leaders did act) after the warning.
Discussion
- Should frontier AI be slowed for safety? Why or why not?
- Do export controls make AI safer or fuel rivalry?
- Where could the US and China realistically cooperate on AI?
- Does centralised control help or harm AI safety?
Vocabulary, reading, and exercises with instant feedback
Spotted a mistake in this lesson? Tell us.