U.S. Lifts Export Controls on Anthropic’s AI Models Amid Growing Chinese Competition

In June 2026 the U.S. Department of Commerce ordered Anthropic to suspend access to its Claude Fable 5 and Claude Mythos 5 models for all foreign nationals, citing national-security concerns. After a two-week shutdown, Commerce Secretary Howard Lutnick issued a letter on June 30 lifting the export-control directive, permitting Anthropic to restore access for “trusted partners” and federal agencies. The company announced it would begin reinstating Fable 5 the following day.

New York Times

President Donald Trump has framed AI development as a global race, urging a “pro-innovation, pro-infrastructure, pro-energy, and pro-export” stance. Simultaneously, his administration has maintained strict export controls on advanced AI chips (e.g., Nvidia and AMD) and banned the use of Huawei equipment, aiming to keep cutting-edge AI technology from Chinese hands. The Anthropic restrictions were part of this broader policy framework.

CNBC

• June 12, 2026 – Commerce Department issues export-control directive to Anthropic.

New York Times

• June 19–30, 2026 – Anthropic negotiates with the White House; Tom Brown leads talks.

CNBC

• June 30, 2026 – Lutnick’s letter lifts controls; Anthropic announces restoration of Fable 5.

CNBC

• Earlier 2026 – Chinese firms release open-weight models: Zhipu’s GLM 5.2, Moonshot AI’s Kimi 2.7, Alibaba’s Qwen 3.

CNBC

• March 2026 – Pentagon labels Anthropic an “unacceptable supply-chain risk,” prompting a lawsuit.

New York Times

• Howard Lutnick – U.S. Commerce Secretary, author of the lifting letter.

New York Times

• Dario Amodei – Anthropic co-founder and CEO, previously targeted for AI-safety views.

CNBC

• Tom Brown – Anthropic co-founder who led negotiations with the administration.

New York Times

• David Sacks – Former Trump crypto and AI czar, vocal critic of Anthropic’s safety stance.

CNBC

• Marc Andreessen – Venture capitalist commenting on Chinese model performance.

CNBC

• Sam Bresnick – Georgetown Center for Security and Emerging Technology researcher.

CNBC

• Flo Crivello – CEO of AI startup Lindy, switched to Chinese models for cost reasons.

CNBC

• Travis Lanham – Co-founder of AI-security startup Armadin, testing Chinese models.

CNBC

• Hed Kovetz – CEO of cybersecurity firm Silverfort, warning of imminent threats.

CNBC

• Zhipu’s GLM 5.2 matches Anthropic’s capabilities on several cyber-benchmark tests while costing roughly ¼ the price per token (Christopher Wood, Jefferies).

CNBC

• Lindy’s migration to DeepSeek’s open-weight models “crashed” its AI cost curve, according to Flo Crivello.

CNBC

• Coinbase reports a ? 50 % reduction in AI spending after adopting GLM 5.2 and Kimi 2.7.

CNBC

• Congressional testimony noted the administration has canceled or frozen > 7,800 research grants while China has overtaken the U.S. in total R&D spending.

South China Morning Post

The lifting of controls restores Anthropic’s ability to compete with rapidly advancing Chinese models that are both capable and inexpensive. Industry leaders warn that without domestic access to frontier models, U.S. firms may lag in cybersecurity, where Chinese open-weight models can already automate reconnaissance and exploit generation. The episode also highlights tension between national-security restrictions and market-driven innovation.

CNBC

The Commerce Department emphasized that Anthropic “has taken steps in close coordination with the U.S. government to address the risks associated with Claude Mythos 5 and Claude Fable 5.” Anthropic thanked users for patience and pledged to redeploy the models. The Pentagon’s earlier designation of Anthropic as a supply-chain risk remains under legal challenge.

New York TimesCNBCCNBCSouth China Morning Post

Tech executives and investors described the export controls as a “gift to China,” arguing they grant Chinese developers valuable time to close the gap. Sam Bresnick called the situation “a pretty good wake-up call.” Flo Crivello’s cost-cutting move and Hed Kovetz’s warning that Chinese models could soon run entire cyber-operations underscore industry concerns. Lawmakers at a House Energy and Commerce hearing criticized the administration for freezing research grants and lagging behind China’s R&D spending.

CNBC

• “Many smart people/AI insiders are saying GLM-5.2 is the first Chinese AI model to match and often beat the American big lab public AI models with no compromises,” — Marc Andreessen, Venture Capitalist

CNBC

• “a pretty good wake-up call,” — Sam Bresnick, Research Fellow, Georgetown Center for Security and Emerging Technology

CNBC

• “We did it, and you could see that cost curve go down, like, crash to the ground,” — Flo Crivello, CEO, Lindy

CNBC

• “If the U.S. government does not let the industry take advantage of this opportunity to get ready, then when the Chinese models reach a similar level, no one will be prepared,” — Hed Kovetz, CEO, Silverfort

CNBC

• “Anthropic has taken steps in close coordination with the U.S. government to address the risks associated with Claude Mythos 5 and Claude Fable 5,” — Howard Lutnick, U.S. Commerce Secretary

New York Times

• “won't take that long.” — Jie Tang, Founder, Zhipu

CNBC

Sources differ on the precise timing of model reinstatement: Lutnick’s letter cites “Wednesday” for Fable 5 restoration, while other reports reference “Friday” as the date of the announcement. Additionally, the U.S. policy framework for regulating open-weight Chinese models remains undefined, leaving a gap in guidance for domestic firms.

CNBC

Congressional hearings are urging faster, coordinated AI investment, while over 1,700 AI bills have been introduced across states, indicating a fragmented regulatory landscape. Industry observers expect further scrutiny of Chinese open-weight models and potential new safeguards before broader deployment. The balance between national-security controls and competitive innovation will shape the U.S. AI sector’s trajectory in the coming months.

CNBC