The General-Purpose AI Code of Practice Officially Released​

On July 10, 2025, the European Commission launched the General-Purpose AI Code of Practice, a voluntary framework developed by independent experts through a multi-stakeholder process. Designed to assist providers of general-purpose AI (GPAI) models in efficiently complying with obligations under the EU Artificial Intelligence Act (AI Act), the Code reduces administrative burdens and enhances legal certainty, balancing innovation with regulatory compliance. The European Commission and AI Board have confirmed the Code’s adequacy as a voluntary compliance tool, allowing signatories to demonstrate adherence to the AI Act without resorting to complex self-certification procedures.

Core Structure: Tiered Governance Framework

The Code comprises three distinct chapters, establishing a risk-based approach:

​​Transparency Requirements​​ (Chapter 1):

Mandatory for all GPAI providers, including disclosure of training data sources, technical limitations, and copyright compliance measures to fulfill Article 53 of the AI Act.

Copyright Safeguards​​ (Chapter 2):

Applies universally, emphasizing intellectual property protection in data usage.

​​Safety and Security Protocols​​ (Chapter 3):

Targets GPAI models with systemic risks (e.g., highly advanced systems), requiring rigorous risk assessments, adversarial testing, and cybersecurity measures under Article 55.

Industry Adoption & Strategic Signatories

As of August 4, 2025, 32 global technology leaders—including Google, Microsoft, Amazon, OpenAI, Anthropic, and Mistral AI—have signed the Code, signaling broad industry endorsement. Notably, xAI (founded by Elon Musk) opted for selective adherence, signing only Chapter 3 (Safety and Security) while committing to demonstrate transparency and copyright compliance through alternative means.

Practical Impact: Triple Benefits of Voluntary Compliance

The Code’s "voluntary compliance alternative mechanism" delivers:

​​Cost Reduction​​: Eliminates resource-intensive self-assessment processes.
​​Enhanced Transparency​​: Data provenance disclosures protect content creators and prevent copyright infringement.

​​Systemic Risk Mitigation​​: "Safety guardrails" for frontier models ensure societal security.

Signatories may join via the official form submitted to EU-AIOFFICE-CODE-SIGNATURES@ec.europa.eu, supported by detailed guidelines and Q&A resources.

Global Implications

The Code sets a new benchmark for global AI governance, harmonizing innovation incentives with regulatory oversight. Its flexible voluntary framework promotes responsible, transparent, and secure AI development, offering a replicable model for other jurisdictions.

Original link : The General-Purpose AI Code of Practice