GPAI Code of Practice
Voluntary EU compliance framework for general-purpose AI model providers, operationalising the EU AI Act's GPAI obligations ahead of 2 August 2026 enforcement.
The GPAI Code of Practice is the EU's voluntary compliance route for general-purpose AI model providers, whose Article 101 fining power, up to 3% of worldwide turnover or €15,000,000, took effect on 2 August 2026.
Last refreshed: 4 August 2026 · Appears in 1 active topic
Which AI companies signed the GPAI Code of Practice and what does non-signing mean?
Timeline for GPAI Code of Practice
Two EU clocks strike on 22 July
Media's AI PivotFinalised on 10 June 2026; opens for signatory sign-up; applies from 2 August 2026 for new AI systems
Media's AI Pivot: EU finalises its AI content-marking CodeBrussels moves Article 50 to December
Media's AI PivotBackground
The GPAI Code of Practice became the EU AI Office's primary lever for enforcing the AI Act's general-purpose model obligations when the Act's Article 101 fining powers took effect on 2 August 2026. Article 111(3) grandfathers models already on the EU market before 2 August 2025 until 2 August 2027, and the AI Office has signalled a collaborative, risk-based approach to that transition rather than immediate penalties.
The Code covers the structural obligations placed on providers of general-purpose AI models: technical documentation, systemic-risk evaluation, incident reporting and cooperation with the AI Office. It is a separate instrument from the Code of Practice on Transparency of AI-generated Content, the Article 50 framework governing machine-readable marking of synthetic media; the two address different obligations, under different articles, for different categories of provider, and should not be treated as one framework.