AI-generated videos and photos used for political disinformation are the scourge of a busy global election year, and Brazil is scrambling to regulate the technology ahead of municipal polls.

In a country of 203 million, which counts more phones than people, Brazilian authorities last week banned the use of deepfake technology and set out guidelines for the use of AI for electoral purposes.

“Video montages can be used to manipulate public opinion, to defame individuals, or to interfere in the democratic process,” Ana Carolina da Hora, a computer specialist at the PUC Catholic University in Rio de Janeiro, told AFP.

The rapid development of AI, turbocharged by the 2022 launch of ChatGPT, has shaken up the online landscape while sparking awe and fear over the future of the technology.

In a video circulating on social media in Brazil, the country’s biggest pop star Anitta and football icon Neymar promote an online gambling scheme, or rather, ultra-realistic deepfakes of the celebrities do.

But in a country hard hit by political disinformation, authorities are particularly alarmed by cases like that of a mayor whose voice was cloned to create an audio file shared on social media in which he insults teachers in his municipality.

Similar cases are being investigated in two other states.

The Superior Electoral Tribunal (TSE) has decided to act. Last week, the use of deepfake technology was formally banned in the October municipal electoral campaign.

Any other type of use of artificial intelligence for electoral purposes must be accompanied by a notice clearly identifiable to the public.

Candidates caught using deepfake technology in the campaign could be blocked from running or have their mandates rescinded if elected.

TSE president Alexandre de Moraes said these were some of “the most modern standards in the world in relation to combating disinformation, fake news and the illicit use of artificial intelligence.”

He warned that deepfake technology could “change the result of the elections.”