Trump administration dismisses AI safety fears while China calls fears 'fearmongering', yet China's spy chief sounds alarm over AI and chip competition.
I've cleaned and tightened this article, but I need to flag a factual concern before you publish: **The article names two AI models that don't exist as of my knowledge—"Anthropic's Claude Mythos" and "OpenAI's GPT-5.5-Cyber."**
Anthropic's actual models are Claude (1-3.5), and OpenAI's are GPT-4, GPT-4o, etc. This may be a translation error from the original Chinese source, or a Livemint error. Please verify with the original source or contact Livemint before running this, as publishing non-existent model names undermines credibility.
Here's the cleaned text:
---
China has dismissed US warnings about artificial intelligence as "fearmongering," yet almost simultaneously, Beijing's top intelligence official issued one of the clearest warnings from the Chinese government about AI's potential to undermine political stability and threaten the Communist Party's hold on power.
Chen Yixin, head of China's Ministry of State Security, called for tighter government oversight of AI, warning that rapidly advancing technology could be exploited for disinformation, cyberattacks, espionage and military competition. His assessment places political security at the centre of Beijing's AI concerns, rather than focusing solely on whether advanced AI could eventually escape human control.
The timing is notable. The warnings come as US and Chinese officials prepare for discussions on AI safety ahead of a planned September 24 meeting between President Donald Trump and Chinese President Xi Jinping in Washington.
In an article published in the state-run magazine *China Cyberspace*, Chen argued that AI could be used by foreign actors to manipulate information and destabilise Chinese society. One concern is deepfake videos, synthetic audio and AI-generated material that could spread rapidly online, creating confusion and undermining public confidence in senior officials.
Chen's warning reflects a broader concern within Beijing that AI could lower the cost of producing and distributing political misinformation at scale. Chinese authorities increasingly treat information and technology control as national security matters.
Chen also warned that advances in AI could make sophisticated cyberattacks easier and cheaper to execute. He specifically pointed to certain US-developed AI models as examples of technologies creating new cybersecurity risks. The Chinese security chief raised additional concerns about foreign intelligence agencies using AI for large-scale espionage, including attempts to obtain sensitive government, military and corporate information. Beijing also worries that Chinese users entering sensitive data into foreign AI systems exposes critical information.
China views AI as a military battleground. Chen argued that AI could transform military operations by improving sensing, targeting and decision-making, citing the use of AI in regional conflicts as evidence of how rapidly military technology is evolving. China treats AI as a strategic technology that could determine its ability to compete with the US—creating a dual approach: faster AI development to strengthen economic and military position, combined with tighter controls over associated risks.
Beijing's response to warnings from US technology leaders differs sharply. Anthropic chief executive Dario Amodei argued that democracies should maintain their technological lead over China and backed restrictions on advanced AI chip exports, calling for measures to slow China's AI progress. China's Foreign Ministry rejected that approach, with spokesman Guo Jiakun characterizing "threat narratives" and "malicious competition" as obstacles to international cooperation on AI regulation. The Communist Party-run *Global Times* criticized Amodei's proposal to contain China's AI development as an "unrealistic delusion."
The competing messages reveal complicated politics surrounding AI. Washington debates whether rapidly advancing AI poses risks ranging from cyberattacks to loss of human control. Beijing rejects attempts to use those concerns to justify restrictions on Chinese technology. Yet China's security leadership warns that AI could be used against the country's political system, critical infrastructure and national interests.
For Beijing, the question is not simply whether AI is dangerous, but who controls the technology, who controls the data and whether rival powers can use increasingly capable AI systems against China's security interests. That tension is likely central to coming US-China discussions, where AI has become both a technology race and a national security issue.