AI was used in 18% of the verified disinformation cases in the EU and Norway in July

253 of the 1,397 fact-checking materials published in July targeted false content created or manipulated with artificial intelligence, according to the European Digital Media Observatory. The proportion increased again after a decline in the spring and approached the record of 20% recorded in January and March.

Artificial intelligence was used for the creation or manipulation of content in 253 of the 1,397 fact-checking materials published in July by organizations in the European Digital Media Observatory (EDMO) network from the European Union and Norway. The proportion reached 18%, after two consecutive months of increase, and approached the record of 20% recorded in January and March 2026.

In short

253 of the 1,397 fact-checking materials analyzed in July involved false content created or manipulated with AI.

These cases represented 18% of the total, compared to lower levels recorded in April and May.

The proportion approached the record of 20% observed in January and March 2026.

Among the examples identified in July is an AI-generated image of Ukrainian President Volodymyr Zelensky and manipulated materials targeting Czech President Petr Pavel and First Lady Eva Pavlová.

AI was also used to fabricate a false video with anti-LGBTQ+ content.

The use of artificial intelligence for producing and manipulating false information has returned to an upward trend this summer, after declines recorded in April and May. EDMO data shows an increase in June and a new rise in July, when the proportion reached 18% of the verified materials.

The level is very close to the maximum of 20% reached in January and March 2026. The monthly evolution shows that AI has become a frequently used technique in cases of misinformation that reach fact-checking organizations.

In July, the 30 organizations in the EDMO network that contributed to the analysis for the European Union and Norway published a total of 1,397 verification materials. In 253 of these, artificial intelligence had been used to create or modify the analyzed false material.

The examples identified by EDMO show that the technology is used to produce visual materials plausible enough to be presented as images or recordings of real events. One of the cases that circulated in several European countries used an AI-generated image in which Ukrainian President Volodymyr Zelensky appeared behind a bulletproof glass.

In the Czech Republic, content created or manipulated with AI targeted President Petr Pavel and First Lady Eva Pavlová. A modified image was used to falsely claim that the constitutional court would not be independent and would make decisions in favor of the president.

Another false material claimed that Eva Pavlová reacted angrily after the leak of documents that allegedly demonstrated she used public funds to purchase jewelry worth four million Czech crowns. The claim was transmitted through a fabricated video.

AI was also used to produce a video with anti-LGBTQ+ content, in which a dog appeared attacking a woman who was said to identify as "female dog." EDMO includes the case among examples of artificially generated content that circulated after the wave of anti-LGBTQ+ materials during Pride.

The cases from July show several ways in which the same technology can be used: complete generation of an image, modification of existing visual material, or making a video that presents as real a situation that did not occur.

The 18% indicator must be reported to the set of verified materials by the organizations participating in the EDMO network. It does not mean that 18% of all content published on the internet in Europe is created or manipulated with AI, but that this technique was identified in 253 of the 1,397 verified cases in July.

The European Digital Media Observatory monitors monthly trends identified by its fact-checking organizations. The report covers the period from July 1 to 31, 2026.

The evolution this year shows significant fluctuations: the use of AI reached 20% of verified cases in January and March, decreased in April and May, then began to rise again in June and July.

At 18%, the level in July is again close to the maximum observed so far in 2026.

https://2eu.brussels/ro/news/ai-a-fost-folosit-in-18-dintre-cazurile-de-dezinformare-verificate-in-ue-si-norvegia-in-iulie