Xi Sells 'Open' AI. His Hackers Already Used Ours Against Us.
Beijing claims it erased America's AI lead. The claims deserve a discount — and Washington is squandering the one advantage China can't copy.
Xi Jinping declared this week that artificial intelligence "should not be a solo performance by a single country" — the autocrat's newest costume: openness. His timing was chosen. Moonshot's Kimi K3 open-weight release had American analysts writing that China "just erased America's AI lead," and Axios reports the race has split in two: Chinese labs giving models away as US labs seal theirs behind APIs.
The benchmark claims deserve a discount. This is the same state whose hackers ran the first AI-orchestrated espionage campaign in history — through Claude, an American model, last fall. Anthropic's own investigation found the group it designates GTG-1002 hit roughly thirty targets with the machine executing up to ninety percent of the work. Beijing preaches openness in public and picks locks in private. The same week Xi praised shared AI, Shanghai surgeons implanted the first commercially approved brain-computer interface, and Beijing moved to restrict AI companions because the machines were suppressing birth rates — the autocrat's version of alignment.
