The consultation is intended to contribute to the development of data protection-compliant approaches for dealing with personal data in the context of large language models (LLMs).

The regulator says it is concerned about LLMs reproducing personal information verbatim from their training data, as well as analogous reproductions which may still identify someone.

Professor Louisa Specht-Riemenschneider, the German data protection commissioner, said: “AI models are revolutionising numerous industries and simultaneously present us with major challenges in terms of transparency, security and data protection.”

The consultation exercise aims to ensure that practical knowledge about AI models is available for regulatory assessment.

“I want to avoid unrealistic and unclear regulations — but for that, I rely on the support of providers and researchers,” the commissioner said.

The consultation is available in English and responses will be accepted until 31 August 2025.