As artificial intelligence advances at breakneck speed across the globe, the European Union is stepping forward with one of the most ambitious regulatory frameworks to date. Alongside the landmark AI Act—the world’s first comprehensive law on artificial intelligence—Brussels has now introduced a voluntary but potentially influential Code of Conduct aimed specifically at developers and providers of general-purpose AI models (GPAI). These are the very systems—like ChatGPT, Gemini, or Claude—that are shaping the digital future.
The timing of this Code is strategic. From August 2025, new transparency and safety requirements for GPAI models will come into effect under the AI Act, with transitional periods for existing systems extending into 2026 and 2027. The Code serves as a bridge between regulatory intent and practical implementation. Those who sign up commit to disclosing technical details on training data, model capabilities, and potential risks. Copyright is another major issue addressed: developers must show how they handle copyrighted material, in response to growing tensions over generative AI and intellectual property rights.
Models deemed to pose “systemic risks”—AI systems with broad and potentially far-reaching societal or economic implications—face additional scrutiny under the Code. This includes independent audits, detailed risk assessments, and safeguards against misuse. The goal is not just compliance but trust: the Code seeks to ensure that AI used in high-impact sectors such as healthcare, law, or education is accountable and transparent.
What sets the Code apart is not only its content but also the way it was developed. More than 1,000 stakeholders—from industry, academia, civil society, and public institutions—contributed to its design. This multi-stakeholder approach is a clear signal that regulation in the digital age cannot be top-down alone; it must be collaborative, flexible, and open to technological evolution. The European AI Office will oversee the implementation and ongoing refinement of the Code, ensuring it adapts as the field progresses.
That said, the Code is not without its critics. Proponents hail it as a milestone for ethical, trustworthy AI. Detractors worry it could become yet another layer of bureaucracy that stifles innovation, particularly for start-ups and smaller tech firms. Some argue that, without legal force, the Code may lack real impact. But the EU is clearly attempting to chart a third path—somewhere between the market-driven model of the United States and the state-led approach of China.
Importantly, the Code does not replace legal obligations under the AI Act; it complements them. It offers guidance and legal certainty, helping businesses to meet compliance requirements ahead of binding technical standards. Early adopters could find themselves better positioned, both legally and reputationally. In an age where trust in digital systems is fast becoming a competitive edge, responsible AI development is not just a legal necessity—it’s a strategic asset.
Ultimately, the EU’s Code of Conduct is a foundational step towards a new kind of digital ethics. It shows that regulation need not hinder innovation—when done transparently, consultatively, and with technological nuance, it can serve as a catalyst. In the months ahead, all eyes will be on the industry’s response. But one thing is already clear: Europe has laid the groundwork for a new relationship between technology, law, and responsibility.
