The upcoming implementation of the EU AI Act in August 2026 signals a shift for companies using generative AI. Article 50 of the Act requires that AI-generated content be detectable and marked, but it does not provide specific guidance on how to achieve these requirements. To fill this gap, the European Commission’s AI Office has developed a Transparency Code of Practice to offer actionable insights for compliance.
In November 2025, the AI Office began creating this Code, aiming to clarify the obligations under Article 50. The Code includes recommendations on technical solutions, such as digitally signed metadata and watermarking techniques, along with labeling formats like standardized EU icons and audio disclaimers. These measures are essential for companies preparing to meet their obligations before the 2026 deadline, as they provide a clearer roadmap for implementation.
Drafted by six independent academic leaders and incorporating feedback from various stakeholders, the Code is currently in its second draft. A final version is expected to be submitted to the European Commission by June 2026. If approved, it will serve as the primary reference for AI transparency compliance across the EU.
Compliance Implications of the Transparency Code
Participation in the Transparency Code is voluntary, but its implications could be significant. If the Commission endorses the Code through an implementing act, following its guidelines may offer a presumption of compliance for participating companies. Conversely, those choosing alternative marking and labeling methods could face increased scrutiny and a heavier burden of proof. They will need to show that their methods effectively meet the criteria outlined in Article 50, including interoperability and reliability.
The division of responsibilities between providers and deployers of AI technology adds another layer of complexity. Providers, such as OpenAI with its ChatGPT and DALL·E models, must embed machine-readable marks in AI outputs and offer free access to detection tools. Meanwhile, deployers—companies that utilize these AI systems—must makes sure that end-users are informed about the AI origins of generated content. This dual responsibility could create operational challenges, particularly for companies that fulfill both roles.
Navigating Dual Roles in the AI Ecosystem
The complications increase for firms acting as both providers and deployers. Major platforms like Meta and TikTok not only develop AI models but also host large amounts of user-generated content that may include AI-generated elements. These companies must balance embedding compliant marks in their own AI outputs while making sure that third-party content meets the transparency obligations set by the Code.
As firms work through these requirements, they must also comply with existing regulations, including the Digital Services Act (DSA), General Data Protection Regulation (GDPR), and various consumer protection laws. The interplay of these regulations with the forthcoming Transparency Code highlights the complex regulatory environment that companies must navigate as they prepare for the AI Act's implementation.
The path ahead for companies in the AI sector will be challenging. As the EU aims to enforce strict transparency in AI-generated content, firms will need to invest in compliance frameworks, staff training, and thorough documentation. The stakes are high, as non-compliance could result in significant legal consequences and reputational damage. Stakeholders will closely watch the development and finalization of the Transparency Code as it shapes the future of AI governance in Europe.
