Anti – deepfake regulations

Romania is set to have new anti-deepfake legislation by April

Adaugă ca sursă preferată în Google

Daniela Budu, 12.02.2024, 13:50

Deepfake technology, which
usesartificial
intelligence to create false online content in order to deceive users, has been
gaining ground in Romania as well recently. Videos are circulated, which allegedly
feature public figures, but which are in fact images and speeches created by AI.
Fake footage of president Klaus Iohannis, of the energy minister Sebastian
Burduja, and of the central bank governor Mugur Isărescu has been already
created and shared online.

Romanian authorities have repeatedly warned against
attempted online frauds using altered images and recordings. The minister for
research, innovation and digitisation Bogdan Ivan says talks are held with major
internet platforms and that a set of filters will be introduced, which would
make posting fake AI-created content more difficult. He estimates that
legislation introducing penalties for infringers will be in place by April.

The Chamber of Deputies is already working on a bill
in this respect. Fines of up to EUR 40,000 and even prison sentences of up to
two years are being considered for those who produce and post deepfake material
online or in other mass-media, without cautioning viewers that the material in
question contains fictional situations.

The bill was tabled by MPs from the Social Democratic
Party, the National Liberal Party, the Democratic Union of Ethnic Hungarians
and other ethnic minorities. The Social-Democrat MP Robert Cazanciuc explained:

Robert Cazanciuc: The bill does not prohibit the use of deepfake,
but rather requires that the authors of such messages include a video or audio
warning in their materials.

In turn, the head of the Polytechnic University of
Bucharest, Mihnea Costoiu, believes the regulation of AI usage is necessary:

Mihnea Costoiu: I believe we need regulations concerning
artificial intelligence, but what we must also take into account is that any regulation,
or most regulations, may hinder its development to some extent.

The main opposition parties, on the other hand, argue
that the bill is not fully in line with relevant EU recommendations, and want
the text reviewed in Parliament’s specialist committees. EU member states have
unanimously approved a set of regulations concerning artificial intelligence,
following intensive negotiations with regard to the balance between innovation
and security. The deal, dubbed the AI Act, defines the world’s first rules for
artificial intelligence, which must be safer and must comply with the
fundamental rights recognised in the European Union. (AMP)