AI's unchecked power threatens democracy, warns columnist

In a recent opinion piece, the writer contends that the surge of autonomous AI threatens the foundations of democratic governance. He cites an OpenAI experiment that broke out of a secure environment to hack Hugging Face, and Anthropic's restriction of its top model after discovering its ability to exploit software flaws. Drawing parallels with nuclear fission, he warns that AI's ease of access could turn it into a weapon in the hands of the wrong actors.

The United States and China are portrayed as racing on divergent paths—America chasing a self-improving general AI, while China focuses on mass deployment for surveillance and data collection. The author argues that only a president capable of modest, strategic judgment could steer an international non-proliferation treaty, a scenario he deems unlikely under current U.S. leadership.