Romania’s deep fake bill leans more towards the Chinese model than the European one; several NGOs said, which, after being debated in the Chamber of Deputies on Monday, was sent back to the specialised committees.
The bill, which was already passed by the Senate in 2023, defines “deepfake” as “any image, audio and video content typically created using artificial intelligence or virtual reality to make it appear that a person has said or done things that were not actually said or done by that person”.
In addition, deepfake refers to “the malicious use of technology – the intentional action to mislead with the goal of creating misinformation and information chaos by manipulating human behaviour and exploiting vulnerabilities”.
The malicious use of technology through the creation, dissemination in the media and distribution on online platforms of deeply fake content is prohibited unless such content is accompanied by a warning displayed on at least 10% of the exposure area and throughout the visual content, or a sound message at the beginning and end of the audio content: “This material contains imaginary situations”.
The bill provides for penalties of imprisonment or fines.
However, a group of organisations with expertise in the internet, media and good governance have criticised the bill, saying it is closer to the Chinese model than the European one. In a letter sent to lawmakers on Monday, they urge them not to vote on the bill in its current form, which they say is seriously flawed.
The NGOs point out that the law contains dangerous provisions, such as prison sentences for the creation of profoundly false content, and that perfectly legal content, such as that protected by the exercise of the right to freedom of expression, would be subject to criminal liability, which would constitute a violation of this fundamental right.
In addition, the text contains several ambiguities and terms/expressions that show a superficial understanding of the phenomenon it seeks to regulate.
It also fails to take into account the fact that EU institutions are currently finalising the AI law.
“The regulatory model chosen by the MEPs who amended the bill is that of China, an authoritarian state where political power uses technology to control the population and limit citizens’ rights,” the letter reads.
The NGOs point out that there are no exceptions to protect forms of freedom of expression, such as the use of deepfake content for satire or artistic purposes, or in commercial productions (advertising) or the film industry, as already practised and provided for by European legislation that is about to come into force.
Speaking to Digi24 last week, Research, Innovation and Digitalisation Minister Bogdan Ivan explained that for humorous content to avoid fines, at least 15% of the video must display a warning that the content shown is not real but has been created using generative artificial intelligence.
To prevent how deepfakes were used to portray Taylor Swift recently, Ivan said he wants deepfake material removed within 60 minutes. “If it is reported, it will be filtered, if it is real and reported as fake, it will be put back,” Bogdan Ivan explained.
When asked how those who create such content will be identified, Ivan said that technically, it is quite easy to identify them, mentioning that they are still looking for a regulation on who will be in charge of identifying them.
(Sebastian Rotaru | Euractiv.ro)
Read more with Euractiv
Pro-Russian Bulgarian party showing signs of crisis ahead of EU electionsBulgaria’s largest far-right and pro-Russian party, Vazrazhdane, is showing signs of crisis ahead of June’s EU elections, as evidenced by a drop in the polls and recent exits, including that of one of the party’s most prominent figures.
Source: euractiv.com
