Glossary
Updated
The AI Office is the central authority responsible for monitoring compliance of GPAI providers, coordinating with national market surveillance authorities, drafting harmonized standards and codes of practice, and maintaining the EU's central database of high-risk AI systems. Established in early 2024, it is the primary regulatory counterparty for any organization providing GPAI models or operating high-risk systems at EU scale.
Example
A foundation model provider whose training compute exceeds 10^25 FLOPs must report safety testing results to the EU AI Office under the GPAI Code of Practice, the Office's first line of regulatory engagement.
