Consultation on the Draft Guidelines on Transparency Obligations Under the AI Act

Why It Matters

The guidance will shape compliance costs for AI firms across Europe and influence global standards for AI transparency, affecting market entry and user trust.

Key Takeaways

• Consultation runs May 8–June 3, 2026, targeting all AI stakeholders
• Draft guidelines clarify labeling duties for interactive and generative AI
• Compliance required by August 2, 2026, with machine‑readable marks
• Deep‑fake disclosures mandated for public‑interest content and biometric systems
• Parallel voluntary Code of Practice supports practical implementation

Pulse Analysis

The European Union’s AI Act represents the most comprehensive legislative framework for artificial intelligence to date, and transparency sits at its core. By requiring providers to disclose when users are interacting with an AI system and to embed machine‑readable markers in generated content, the regime seeks to curb misinformation and protect fundamental rights. The Commission’s draft guidelines, released on May 8, 2026, are the latest step toward operationalising Article 50, translating legal language into actionable standards for developers, startups, and multinational tech firms alike. The initiative also signals the EU’s intent to harmonise AI governance across member states.

The proposed rules impose concrete technical obligations. Interactive systems must display clear notices, while generative models—especially those capable of producing deep‑fakes or synthetic media—must embed metadata that can be read by detection tools. Deployers of AI that influence public opinion, emotion‑recognition, or biometric categorisation will also need to inform users of potential manipulation. Companies will therefore need to invest in labeling pipelines, update user‑interface designs, and possibly redesign model outputs to ensure compliance before the August 2, 2026 enforcement date. Compliance will likely be verified through periodic audits and mandatory reporting to national authorities.

Beyond the EU border, the guidelines are likely to set a de‑facto global benchmark, as many multinational vendors align their products with the strictest standards to avoid fragmented compliance regimes. The parallel voluntary Code of Practice offers a pragmatic pathway for firms that seek early adoption, reducing the risk of costly retrofits. Stakeholders are urged to submit feedback by June 3, 2026, to shape the final text, while simultaneously preparing internal governance, audit mechanisms, and cross‑border data‑handling policies to meet the upcoming transparency obligations. Early alignment can also provide a competitive edge in markets that value ethical AI.