AI finds 10,000 vulnerabilities. China is copying the models.
This title could be clearer and more informative.Try out Clickbait Shieldfor free (5 uses left this month).
Frontier AI models like Anthropic's Claude Mythos (Project Glasswing) have found over 10,000 high-severity vulnerabilities across major operating systems and browsers. The first confirmed case of an AI discovering and weaponizing a zero-day exploit in the wild occurred in May. Meanwhile, Chinese labs including DeepSeek, MiniMax, and Moonshot AI have conducted industrial-scale distillation campaigns against US frontier models — generating 16 million total exchanges via fraudulent accounts to replicate offensive capabilities. The US policy response is a voluntary 30-day pre-release review with no enforcement power, weakened from an original 90-day proposal. The core dilemma: restricting access slows defensive tool diffusion to allies, while open access accelerates offensive capability transfer to adversaries.
