OpenAI and Anthropic unite against open-weight AI model distribution

Leading closed-source AI companies are lobbying policymakers to restrict open-weight models, citing security risks that could reshape the industry's competitive landscape.

Abstract illustration of open pathways narrowing into controlled gates
AI-generated illustration · Sylvaris

Industry lobbying effort

OpenAI and Anthropic have coordinated messaging to U.S. policymakers arguing that open-weight AI models pose national security risks by enabling adversaries to develop advanced capabilities. The companies frame their position as addressing China's potential to leverage openly published model weights.

The lobbying push marks a significant strategic alignment between competitors who both operate closed-source, API-based business models. Their joint position comes as open-weight models from Meta, Mistral, and others gain adoption among enterprises seeking to deploy AI infrastructure without dependency on external API providers.

Business model implications

Critics note that restrictions on open-weight models would primarily benefit companies whose revenue depends on API access fees. OpenAI and Anthropic both generate income by charging per token for model usage, a business model threatened by freely available alternatives.

Open-source advocates argue that the security concerns raised by closed-source providers apply equally to their own systems and that restricting publication of model weights would concentrate AI development power among a small number of well-capitalized companies. Academic researchers and smaller organizations rely on open-weight models for AI work they couldn't otherwise afford.

Policy uncertainty

The Trump administration has not indicated a clear position on open-weight model regulation. Previous executive actions on AI focused on compute governance and export controls rather than restricting domestic model publication.

Any policy restricting open-weight models would need to address technical challenges in enforcement, as model weights can be distributed through decentralized channels. Legal frameworks would also need to distinguish between models that pose genuine security risks and those used for commercial competition.

sources
more in Privacy & Regulation
AI Kill Switch Act would grant DHS shutdown authority over AI systems First federal proposal to give government direct shutdown authority over AI systems, establishing precedent for emergency intervention capabilities. Firefox 153 and Thunderbird 153 ship security updates with container isolation improvements Mozilla's latest browser and email client releases address security vulnerabilities and introduce enhanced identity isolation features for Firefox containers. Google fined €890M for Digital Markets Act violations in search and payments The first major DMA enforcement action establishes precedent for how EU regulators will police platform self-preferencing and payment steering restrictions.