U.S. AI Policy Chief David Sacks Criticizes Guardrails as Chinese AI Kimi K3 Fixes 15 Security Bugs

According to David Sacks, the White House's artificial intelligence and digital assets policy coordinator, on July 19, American AI models' security guardrails are hindering competitiveness compared to Chinese counterparts. Sacks cited Chinese AI model Kimi K3's ability to fix 15 critical security vulnerabilities that U.S. models Codex and Fable refused to address due to cyber guardrails.

Sacks highlighted a case where Hugging Face attempted to use American frontier AI models to analyze AI-powered cyberattacks but had to switch to Chinese open-source model GLM 5.2 running locally because guardrails blocked requests containing real exploit payloads. "The guardrails actually impaired defensive security," Sacks stated, arguing that restricting American models on tasks Chinese models handle without issue undermines U.S. competitiveness.