Representatives from the AI company Anthropic sat down with U.S. Commerce Department officials this month to talk about potential changes to export policy on artificial intelligence technology. The meeting, confirmed by people familiar with the matter, centered on proposed reforms that could reshape how AI models and related hardware flow across borders.
The meeting’s context
The closed-door discussion comes as the Biden administration reviews existing export controls on advanced semiconductors and AI software. The Commerce Department’s Bureau of Industry and Security has been tightening rules on shipments to China and other rivals, and new reforms could expand those restrictions or carve out exceptions for certain types of AI collaboration.
Anthropic is one of the most prominent names in AI safety research. The company develops large language models and has long argued for careful regulation of the technology. Its presence at the table signals industry interest in shaping rules that could affect how models are trained, tested, and deployed internationally.
What’s at stake
Export policy on AI touches everything from cloud computing access to the sale of chips that power training runs. The current rules already limit sales of high-end Nvidia GPUs to certain countries. Proposed reforms could extend those limits to include the export of AI model weights or impose licensing requirements for cross-border research collaboration.
Anthropic’s focus has been on building safe AI systems. The company has publicly backed government oversight but also warned that overly broad restrictions could hurt innovation. The Commerce Department meeting suggests Anthropic is trying to ensure its voice is heard as the policy takes shape.
Neither side has released details of what was discussed. The department routinely meets with tech companies to gather input on national security and economic competitiveness.
Next steps unclear
The Commerce Department has not set a timeline for any new rule. A public comment period on earlier proposed changes ended in late 2024. Industry watchers expect additional guidance on AI-specific export controls in the coming months, but no firm date has been announced.
For now, Anthropic and other AI firms will likely keep engaging with regulators. The meeting was one of several the department has held this year with major AI companies. What comes of those talks — and how the final policy reads — is still an open question.
