Sacks Says US AI Guardrails Hand China a Competitive Edge

US AI developers use cybersecurity guardrails to prevent frontier models from assisting malicious hacking, but the same controls can block legitimate vulnerability research and incident response. David Sacks, a former White House AI and crypto czar who now co-chairs the President’s Council of Advisors on Science and Technology, has argued for lighter-touch regulation. The dispute matters because restrictions on dual-use coding capabilities could determine whether US labs retain an edge over increasingly capable Chinese rivals.

On July 20, Sacks cited a test showing Beijing-based Moonshot AI’s Kimi K3 patched 15 critical security flaws that OpenAI’s Codex and Anthropic’s Fable refused to handle because of cyber guardrails. He said there was no reason to restrict American models from tasks Chinese systems perform, warning that the policy would only erode US competitiveness. Moonshot launched Kimi K3 on July 16 and said it would release the model’s full weights by July 27.