The risk identified by KNAB is based on the precedents of the USA, Poland, Slovakia and several other countries where artificial intelligence tools have been used in the context of elections. According to the KNAB, it is likely that artificial intelligence could also be used during the pre-election campaigning period of the upcoming European elections.
“Artificial intelligence offers a variety of content-making capabilities. Manipulation typical of the pre-election campaigning period could involve synthesizing the voice of deputy candidates or including portraits in images or videos simulating false conversations or situations,” KNAB said.
While materials made with artificial intelligence tools can be difficult to identify, it is possible to see signs of artificially constructed content, the office noted. Human portraits produced by artificial intelligence, for example, are characterized by unnatural grimaces, unsymmetrical body parts and proportions, and the inconsistency of lip movements with spoken text. Similarly, the content thus created is characterized by phrases that are not typical of the Latvian language, while in conversations – unnatural speech mannerisms.
The pre-election campaigning period before the European elections began on February 10 and will run until election day – June 8 .
Supervision of pre-election campaigning shall be performed by the KNAB, the National Electronic Mass Media Council (NEPLP) and the State Language Center. The CPCB monitors restrictions and prohibitions on pre-election campaigning in relation to pre-election expenditure limits, use of administrative resources, and covert pre-election campaigning on the Internet, including social networks, public places, as well as in press publications. The NEPLP, on the other hand, monitors pre-election campaigning on television and radio and the State Language Centre – the use of the official language during the pre-election period.