Breaking

White House AI Advisers Clash Over Regulatory Strategy Against China's Kimi K3 Open Model

July 19, 2026 at 08:57 EDT

Policy & Regulation
Foundation Models
Open Source

A public dispute has erupted among senior figures in the Trump administration's technology orbit over whether the U.S. should deliberately create "regulatory risk" to steer companies away from Chinese open-weight AI models, following the release of Moonshot AI's Kimi K3. The exchange pits Dean W. Ball, OpenAI's Head of Strategic Futures and a former OSTP adviser, against David Sacks, the Craft Ventures co-founder who co-chairs the President's Council of Advisers on Science and Technology.

July 17, 2026 · U.S. AI Policy

Manufacture "Regulatory Risk" to Block Chinese AI? Two Trump-Orbit Advisers Clash

The release of Moonshot AI's Kimi K3 — billed as the world's largest open-weight model — has split senior figures in the administration's tech orbit over whether the U.S. should deliberately create uncertainty to steer firms away from Chinese systems.

2.8T

Parameters — largest open-weight model to date

Token context window for long-horizon work

MoE

Sparse mixture-of-experts, native multimodality

Where Kimi K3 Ranks (overall capability)

Claude Fable 5

frontier (lead)

GPT-5.6 Sol

frontier (lead)

Kimi K3

just below frontier

Other models

tested field

Takeaway: Kimi K3 sits below the two U.S. frontier leaders overall, yet consistently ahead of the rest of the tested field — with agentic coding roughly on par with early-2026 top public models.

Dean W. Ball · OpenAI

Suggests the administration could counter Chinese open-weight models by generating "regulatory risk" via soft-law guidance across agencies — discouraging enterprises from adopting them. Warns of a "decelerationist" open-source strategy.

David Sacks · PCAST

Calls the "weaponization of regulatory uncertainty" as a competitive tool "completely unacceptable." Regulation should rest on facts, logic and evidence — not fear, uncertainty and doubt used to shield incumbents.

The open question for U.S. AI policy

Will the response to China's open models be grounded in evidence-based rules — or in deliberately manufactured uncertainty?

Backers cite: frontier-level agentic coding, open competition against a closed-lab duopoly.

Critics cite: national-security risk, possible backdoors, higher "token-hungry" inference costs.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.