China’s Spy Agency Warns AI Threatens National Security

China’s intelligence service is warning that artificial intelligence is creating new national-security threats through deepfakes, automated propaganda, data leakage and foreign influence operations.

The Ministry of State Security has highlighted the ability of AI to generate convincing false information and operate automated “troll armies” capable of influencing public opinion at scale.

It has also warned about sensitive data leaking through AI systems and foreign actors exploiting the technology to undermine China’s political and ideological stability.

The warning comes as Beijing develops its broader response to increasingly capable AI.

China has separately begun preparing for systems that could replicate themselves, seek power or escape human control.

SAFE FROM WHAT?

China and the West increasingly agree on one thing.

Advanced AI can be dangerous.

But they do not necessarily define the danger in the same way.

Western frontier laboratories such as Anthropic increasingly focus on autonomous systems, cyber capabilities and the possibility of humans losing control of increasingly powerful models.

China’s intelligence establishment is also concerned with something different.

Deepfakes.

Foreign influence.

Political destabilisation.

Loss of state control over information.

Chinese state-backed media has meanwhile attacked Anthropic’s proposal to slow frontier AI development as a potential American “Cold War playbook”.

That exposes a fundamental obstacle to international AI governance.

Washington and Beijing may both demand “AI safety” while trying to protect themselves from very different things.

Before the world’s two AI superpowers can agree on how to make AI safe, they may first have to agree what “safe” actually means.

Sources

Financial Times: China’s intelligence service warns of AI national-security risks

Reuters: Chinese state media attacks Anthropic AI slowdown proposal