A major practical layer of the EU AI Act started applying on 2 August 2026. Article 50 is designed to make artificial intelligence more visible: people should know when they are speaking to a machine, and certain synthetic media must be detectable or visibly disclosed. The rules do not make every AI output untrustworthy. They make provenance part of the product.
- What changed on 2 August
- A label and a machine-readable mark solve different problems
- Not every edit is treated like a deepfake
- What this means for publishers and creators
- The VerdictLab standard
What changed on 2 August
The European Commission says providers of systems that interact directly with people must design them so users are informed that they are dealing with AI, unless that is already obvious. Providers of generative systems must also support machine-readable detection of artificially generated or manipulated output. The same date activated disclosure duties for deployers using emotion recognition, biometric categorisation, deepfakes and some AI-generated public-interest text.
A label and a machine-readable mark solve different problems
A visible notice helps the person in front of the screen. A machine-readable mark is intended to travel with the file so platforms, publishers and verification tools can identify synthetic or altered material. Neither method proves that a claim is true. Together, they give audiences and technical systems more information about origin, editing and responsibility.
Not every edit is treated like a deepfake
The Commission’s guidance describes scope, exceptions and practical examples. Standard editing and assistive functions are not automatically equivalent to deceptive synthetic media. Context matters: a cinematic effect, an AI customer-service agent and a fabricated public statement create different risks. Companies that do not use the voluntary transparency code still need to demonstrate compliance through other adequate measures.
What this means for publishers and creators
Teams should map where AI appears in research, generation, editing, customer support and publication. Disclosures need to be placed where a reader can actually see them, while provenance data should survive export when technically feasible. A generic policy page is useful, but it should not replace an article-level or media-level notice when the content itself falls within a disclosure duty.
The VerdictLab standard
VerdictLab uses AI to accelerate research, synthesis and translation, while a human editor selects sources and decides what is published. Our articles identify that process, link to original sources and do not present generated illustrations as documentary photographs. The new European rules reinforce the principle behind that approach: useful AI should make accountability clearer, not harder to find.
Transparency is not a warning that AI is automatically wrong; it is context that helps people judge what they are seeing.
VerdictLab editorial note
Original sources
This VerdictLab report is an original summary and analysis based on the reporting below.
How this was created: VerdictLab selected and checked the cited reporting, prepared an original summary with AI assistance, and translated it for accessibility. We link to the original publishers and do not reproduce their articles.