Overview
- On July 10 the European Commission released the long-awaited voluntary code of practice to guide compliance with the EU’s AI Act for advanced general-purpose AI models.
- The code mandates transparency in training data, adherence to EU copyright rules and basic security measures before binding obligations begin on August 2.
- Companies can opt into the framework once approved by member states, gaining reduced administrative burdens and increased legal certainty versus alternative compliance routes.
- Binding enforcement for new AI models will start one year after August 2, and existing systems have until August 2027 to meet the requirements.
- Nearly 50 leading European firms have called for a two-year pause on enforcement and watchdogs have accused Big Tech of influencing the code’s drafting process to weaken its provisions.