Skip to content
Listen & Watch
Promptcast ▶▶ in German· 8 September 2025

10.8 Update 1 🇩🇪 The Purpose of the “General Purpose AI Code Of Practice”

by 4 min read

There is already an introduction to the EU's effort to regulate the AI market, protect businesses and consumers and create legal certainty: Episode 10.8 - No risk, no fun with AI? What does the law say?

To make implementation easier, the EU has also published a “Code of Practice for general-purpose AI”. It is meant to ease the step-by-step implementation of the law, and it has the added benefit of making this complex issue easier to understand and defusing the various pitfalls of the “Simplicity Trap”.

It also creates a framework for moving AI out of the rather fuzzy realm of social IT and into the far more compliance-heavy realm of business IT.

The Purpose of the “General Purpose AI Code Of Practice”

(If you belong to the TL;DR generation 😉 and would rather listen to the whole thing as a Promptcast, here's the shortcut -> Promptcast on the topic)

The General Purpose AI Code of Practice (GPAI Code of Practice) is a voluntary set of rules from the European Union designed to help providers of general-purpose AI models meet their obligations under the EU AI Act. Published by the European Commission on 10 July 2025, the document is the world's first comprehensive guide to regulating general-purpose AI models. (1) (2) (3)

Main Purpose and Objectives

The overarching goal of the Code of Practice is to improve the functioning of the internal market and promote the uptake of human-centric, trustworthy AI, while ensuring a high level of protection for health, safety and fundamental rights. Specifically, the Code serves as: (4)

  • A compliance guide for Articles 53 and 55 of the AI Act (5) (1)
  • A bridging solution between the GPAI obligations taking effect (August 2025) and the adoption of harmonised standards (August 2027 or later) (1)
  • An instrument for reducing administrative burden and increasing legal certainty for providers (6)

Structure and Scope

The Code is divided into three main chapters:

1. Transparency Chapter

Applies to all GPAI model providers and covers:

  • Producing comprehensive technical documentation (7) (1)
  • A standardised Model Documentation Form for capturing relevant information consistently (3)
  • Providing information to downstream providers and regulators (8)
  • A public summary of the training data (9)

Also mandatory for all GPAI providers:

  • Implementing policies to comply with EU copyright law (2)
  • Using state-of-the-art technologies to identify and respect copyright (3)
  • Practical solutions for complying with EU copyright law (2)

3. Safety and Security Chapter

Applies only to providers of models with systemic risk (models trained with more than 10²⁵ FLOP of compute): (10) (11)

  • State-of-the-art model evaluations (1)
  • Risk assessment and mitigation (1)
  • Reporting serious incidents (1)
  • Adequate cybersecurity measures (1)

Although the Code is not legally binding, it gives providers a recognised way to demonstrate compliance. The European Commission and the AI Board have confirmed that the Code is an adequate voluntary instrument for meeting the obligations of the AI Act. (12) (7) (6)

Enforcement aspects:

  • From August 2026, the AI Office can impose fines of up to 3% of global annual turnover or 15 million euros (whichever is higher) (13)
  • Signatories of the Code get a grace period for implementation and are regarded as “acting in good faith” (7)
  • Non-signatories have to provide alternative, potentially more burdensome proof of compliance (1)

Development Process and Stakeholder Involvement

The Code emerged from a multi-stakeholder process involving:

  • Nearly 1,000 participants from industry (14) (12)
  • Academic and independent experts (1)
  • GPAI model providers and downstream providers (1)
  • Civil society representatives (1)
  • Representatives of EU member states (14)

Strategic Significance for Digital Transformation

From a compliance and business application perspective, the Code represents a first regulatory framework for global AI governance. It creates:

For companies:

  • Planning certainty for AI investments
  • Standardised compliance procedures
  • Competitive advantages through early adaptation

For the industry:

  • Harmonised minimum standards
  • Less regulatory fragmentation
  • Support for responsible AI innovation

For the EU internal market:

  • Building trust among consumers and businesses
  • Positioning as a leading location for AI regulation
  • A balance between innovation and risk management

EU vs. the Wild West approach

Without the protection and the limits of the law, freedom inevitably turns into a tyranny of the strong over the weak. That is not just one reading of Kant, it's a fact. While the Trump administration conjures up the Wild West again through sweeping deregulation, knowing full well that the US tech giants currently hold the bigger lever, the EU is taking an ethically far better path.

Even if, like almost every entrepreneur, you're no fan of state regulation, you have to admit that the EU approach makes a fairer market possible: it strikes a balance between recommendation and hard regulation, and it is proactive. It protects both consumers and businesses from this overwhelming power and creates better conditions for competition and innovation. How well we actually use that advantage, however, is up to us.


The Promptcast on the topic

It has the advantage of covering the topic in more breadth and detail. So it's worth a listen if you want the fuller picture. Enjoy 🙂

Promptcast · 17:17 · in German
10.8 Update 1 🇩🇪 The Purpose of the "General Purpose AI Code Of Practice"

🚀 Service: Book your own AI expert for free 😉