---
title: "Commission Guidelines Explain AI Act Transparency Duties as the 2 December Date Nears"
url: "https://grc.plus/updates/ai-act-article-50-guidelines-transparency/"
type: "Update"
updated: "2026-11-13"
publisher: "GRC+"
---

# Commission Guidelines Explain AI Act Transparency Duties as the 2 December Date Nears

## What happened

On 20 July 2026 the European Commission [published guidelines](https://digital-strategy.ec.europa.eu/en/news/commission-publishes-guidelines-transparency-obligations-providers-and-deployers-certain-ai-systems) on the transparency obligations in Article 50 of the AI Act. The [document, C(2026) 5054 final](https://ai-act-service-desk.ec.europa.eu/sites/default/files/2026-07/guidelines_on_the_implementation_of_the_transparency_obligations_for_certain_ai_systems_under_article_50_of_the_ai_act_bzptwqhk0ikg1dtlddap41psfy_131215.pdf), is an annex to a Communication approving the content of the draft guidelines. Guidelines assist compliance. They are not a new legal act.

## What it changes

Paragraph 153 says Article 50 applies from 2 August 2026 to all in-scope systems, whenever they were placed on the market.

It also says the AI Omnibus envisages a targeted exception. Providers of generative AI systems placed on the market or put into service before 2 August 2026 get until 2 December 2026 to meet the marking and detection duty in Article 50(2). The disclosure duty for systems that interact directly with people applied from 2 August, including for systems that are partly interactive and partly generative. We rely on the guidelines for this date and did not open the Omnibus text.

Paragraph 154 says outputs created before 2 August 2026 need no retroactive marking or labelling. Public-interest text generated earlier but published on or after that date does need a label.

## Who is affected

Providers of AI systems that interact with people or generate synthetic content. Deployers that expose people to deep fakes, emotion recognition or biometric categorisation, or that publish AI-generated text on matters of public interest without human review or editorial control.

## What to do now

- List your AI systems and record whether you are provider or deployer for each.
- For generative systems already on the market, confirm that machine-readable marking works before 2 December 2026.
- Check that chatbots and voice agents already tell people they are dealing with AI.
- Set a labelling rule for deep fakes and for public-interest text, and record who does the human review.

Principal Consultant

## [Ahmet Ucar](https://grc.plus/consultant/ahmet-ucar/)

Helps clients comply with their regulatory requirements with a plan their board can approve.

[Book a 30-Minute Call](https://meetings-eu1.hubspot.com/ahmet-ucar)

## [AI Governance Officer as-a-Service](https://grc.plus/service/ai-governance-officer-as-a-service/)

A named AI governance officer who keeps your use of AI in line with the EU AI Act: inventory, risk classification, AI literacy and policy.

[View the Service](https://grc.plus/service/ai-governance-officer-as-a-service/)
