What Should Frontier AI Developers Disclose About Internal Deployments?

2026-04-24

AI Summary

This study addresses the critical lack of external oversight when leading AI developers deploy high-capability models internally, underscoring the urgent need for transparent disclosure to substantiate their safety. The work proposes the first systematic framework for structured disclosure, delineating essential information across four dimensions: model capabilities, deployment contexts, safety mitigations, and governance mechanisms. Grounded in policy analysis and risk governance methodologies, the framework clarifies the benefits, limitations, and associated risk-mitigation strategies for each disclosure category. Designed for direct integration into model cards and regulatory reporting, this approach significantly enhances the trustworthiness and regulatory compliance of internally deployed AI systems.

Technology Category

• Philosophy and Ethics of AI: Safety, Robustness & Trustworthiness
• Natural Language Processing: Safety and Robustness
• Humans and AI: Explainable AI (XAI) for Human Understanding

Application Category

• Security and Privacy: Data transparency and provenance
• User Modeling, Personalization and Recommendation: Accountability, Transparency, and Ethics for personalization
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Frontier AI developers are increasingly deploying highly capable models internally to automate AI R&D, but these deployments currently face limited external oversight. It is essential, therefore, that developers provide evidence that internally deployed models are safe. While recent work has highlighted the risks of internal deployments and proposed broad approaches to transparency and governance, there remains little guidance on the specific information developers should disclose about them. We address this gap by identifying key information that companies should disclose about internally deployed models across four categories: capabilities, usage, safety mitigations, and governance. For each category, we analyse the key benefits and limitations of disclosure and consider how disclosure-related risks can be mitigated. Our framework could be used by developers to inform both public transparency documents, such as model system cards, and private periodic reports required under emerging frontier AI regulation.

Problem

• frontier AI
• internal deployment
• disclosure
• AI safety
• transparency

Innovation

• internal deployment
• AI transparency
• safety disclosure
• frontier AI governance
• model system cards