What you will achieve now
A decision or proof applied to “AI Act transparency: audit chatbots and synthetic content”.
- 1 Understand the criterion
- 2 Do a small practice exercise
- 3 Save evidence
A brief note covering what you did, what went well, what failed, and what you would review next.
- Locate chatbots and content generated or manipulated with AI.
- Distinguish information when interacting with an AI, machine-readable marking, and notice to people who are exposed.
- Save a dated audit without presenting the exercise as legal advice.
What the official guidance says
The European Commission’s guidelines on Article 50 explain transparency obligations for certain systems. They apply from 2 August 2026. Among the cases described are informing anyone who interacts directly with an AI, marking generated or manipulated content in a detectable way, and notifying people exposed to deepfakes, emotion recognition systems, or texts on matters of public interest without human review or editorial control.
There is an important nuance: the Commission also describes a transitional period for certain generative systems placed on the market before 2 August 2026, until 2 December 2026. Do not turn a start date into a one-size-fits-all answer: check the system, your company’s role, and the current guidance.
Five-step audit
- List every chatbot, assistant, and public channel.
- Indicate whether your company provides it, deploys it, or only uses it internally.
- Review what content is generated or modified and whether there is editorial review.
- Check the visible notice and technical marking when applicable.
- Save a screenshot, URL, owner, date, and corrective action.
Primary sources
- Commission guidelines on transparency obligations .
- Guidance on transparency of AI-generated content .
- Official summary of transparency rules .
If you have saved the evidence from this lesson, continue with “Diagnosis and first AI pilot for an SME”. If not, repeat the check before moving on.
Learn with a verifiable reference
Aulafy distinguishes stable concepts from data that changes—versions, prices, models, and commands. Check the course card for the review date, verified scope, and primary sources.
“Editorial review” means structure, claims, and sources have been reviewed. It does not mean every command has been executed: when a technical test exists, it will be labelled as such.