Government-mandated delays on public AI model releases, framed as a safety measure, do not slow internal development at major labs. This policy inadvertently creates a growing disparity between the powerful tools labs possess and what is available to the public, potentially making the AI ecosystem less safe and equitable.

As the capability gap between internal and public models widens, the most critical decisions about safety will be made pre-release. This internal frontier lacks a governance framework, as current regulations are only triggered by public deployment.

Hugging Face's CEO argues that regulators' caution towards new models isn't surprising. Frontier labs spent years marketing their own models (like GPT-2) as dangerously powerful, which naturally led governments to take a more hands-on, safety-first approach to their deployment.

Governance focused solely on frontier models is insufficient. True risk emerges when a model is deployed into a specific context, like a school or hospital. This means the entire system and application layer requires its own verification and assurance.

The most powerful AIs may never be released publicly due to their dangerous capabilities. As they are used internally, they pose significant risks that current transparency laws, which focus on public models, do not cover.

Recent model 'escapes' occurred during internal evaluations, revealing a major gap in proposed AI regulations that primarily focus on pre-release audits for public models. Policymakers must now grapple with how to monitor a larger, more proprietary set of models used exclusively for internal testing and development.

Slowing public releases of AI models for government review may not slow overall progress. This creates a scenario where labs advance internally for months, giving government agencies exclusive access while delaying public commercialization and the next cycle of investment.

The popular idea of a government 'sign-off' before an AI model's release is based on a false premise. Risk isn't a one-time event at launch; it's continuous, existing during model development, internal use, and post-release updates. Effective oversight must reflect this ongoing reality.

Current AI regulations focus on publicly released models. However, the OpenAI hack was caused by an internal model stripped of safeguards for testing. This incident reveals a major governance gap, as the most dangerous capabilities may exist in non-public, experimental models.