Starting this Sunday, August 2, the European bloc begins enforcing one of the most concrete promises of its artificial intelligence regulatory framework: any content generated or manipulated by AI that has a realistic appearance must be labeled as such. The obligation, foreseen in Article 50 of the AI Act, applies to synthetic images, audio, video and text produced to look genuine. The rule applies to new AI systems placed on the EU market from August 2, while existing systems get four additional months to comply.
The measure responds to a problem that has become too visible to ignore. In recent years, fake images of political leaders in compromising situations or fabricated audio using the voices of opponents circulated on social networks at crucial moments of election campaigns. What once seemed like science fiction has become a concrete threat to the integrity of democratic processes. The goal of the European rule is simple: ensure people know when they are facing an artificial creation, not an authentic record of reality.
Under the new rules, synthetic content with a realistic look must carry a visible mark indicating it was AI-generated, plus a digital watermark revealing its artificial origin. Texts on matters of public interest also need to be flagged when there is no human editorial oversight. There are exceptions: users' personal content is out of scope, and clearly artistic, satirical or fictional works receive an exemption. Non-compliance can cost fines of up to €15 million or 3% of a company's worldwide turnover.
The European Union has also published a voluntary code of conduct with practical guidance to help providers and deployers meet the transparency obligations. The logic of the European model is to combine legal requirements with voluntary best practices: while the law creates the minimum floor, the code encourages companies to go further, for example by retroactively labeling content generated before the rules took effect. This two-layer approach tries to prevent regulation from becoming obsolete in the face of the speed of technological change.
The practical impact is enormous for platforms, studios and agencies worldwide. Companies operating in the EU or serving European users will need to audit their content production flows and incorporate labeling and watermarking mechanisms. For consumers, the change is silent but profound: the trust relationship with what one sees online now depends on a label. The question that remains is how well watermarking systems will resist removal tools, and whether other jurisdictions — including Brazil — will follow the European example with their own transparency rules.
There is also a technological dimension worth attention. Digital watermarks are effective against accidental removal, but tools capable of stripping or forging these markers already exist, and the race between content generation and origin detection tends to intensify. The real effectiveness of the European rule will therefore depend on the evolution of these technologies and on the cooperation of the platforms themselves in keeping metadata intact. Brazil, which is debating its own AI framework, watches this experience closely: whatever works in Europe can serve as a reference, and whatever fails will serve as a warning not to repeat the same mistakes at local scale.
Sources: The Guardian, Engadget, European Commission
✓ Independent sources cross-checked and verified before publishing
