When AI secrecy meets regulatory science transparency: which way for algorithmic transparency at EFSA and EMA?

The advent of AI in EU regulatory science is a litmus test for transparency. This article examines how EFSA and EMA are responding. It is argued that both agencies are subject to stringent safeguards for AI transparency. However, these lie in the guarantees developed around their scientific assessments (i.e., the logic of Regulatory Science Transparency), rather than in the AI Act. From the vantage point of Regulatory Science Transparency, expert accountability and legitimacy, EFSA’s and EMA’s emerging approaches to AI transparency yield mixed results: only EFSA’s response holds out the prospect for robust democratic governance in the age of AI.