Neural networks were not allowed into campaigning

Neural networks were not allowed into campaigning

The Government Commission on Legislative Activities on Monday gave a negative opinion on the LDPR initiative regarding the use of images or voices generated by artificial intelligence (AI) in election campaigning. The Liberal Democrats wanted some of the restrictions that apply to campaigning to not apply if it is created using neural networks. Experts believe that such bans should not be lifted, and in general there are no special restrictions on the use of AI in the election campaign in the current legislation.

Deputies from the Liberal Democratic Party introduced a bill on the use of AI in election campaigns to the State Duma in October last year. They proposed amendments to Art. 48 of the federal law “On Basic Guarantees of Electoral Rights”, which regulates the rules of campaigning. It was proposed to add a clause according to which “the restrictions and prohibitions established by this article do not apply to the use of... images and (or) voice (synthesized speech) created using artificial intelligence technologies.” Now, for example, images and statements of people can be used in campaigning only with their written consent. Instead, the LDPR proposes that voters must be informed “about the use of this technology.” A similar amendment has been prepared to the law on elections of State Duma deputies.

The bill will allow “to significantly reduce the costs of parties during the election campaign,” the explanatory note says. Earlier, project co-author Yaroslav Nilov gave Kommersant an example of such savings: there will be no need to pay fees to announcers and artists.

He also recalled the “Russian first political algorithm “Zhirinovsky”” created by the LDPR (presented at SPIEF in June 2023) and suggested that such technologies will be used more and more, which means it is necessary to create a legal framework.

The government commission, however, did not appreciate the idea of ​​the LDPR and referred to the legal position of the Constitutional Court, according to which “special regulation of election campaigning is aimed primarily at preventing the possibility of voters forming a distorted idea of ​​the candidate, as well as at ensuring equal rights of candidates and protection of the rights and interests of citizens who are not candidates" (definition of the Constitutional Court of June 27, 2017 No. 1226-O). In addition, the commission's review states that the approach allowing the use of "images of any individuals without their written consent in campaign materials if such an image is created using artificial intelligence technologies" is not consistent with the principle of privacy.

According to experts, it is too early to legalize the use of neural networks in campaigning in this way. “The issue of the limits of the use of AI in elections should be resolved after general legal regulation in this area has been established. For example, labeling synthesized content, with the exception of that created for artistic purposes,” says political scientist Pavel Sklyanchuk. “AI is interesting for parties and political strategists. Augmented reality technologies and neural networks are already being used to create “embellished” images. However, it is in the interests of the voter and society that real politicians are not replaced with their avatars during campaigning, since deepfakes can mislead people, provoke anxiety and reduce the level of digital hygiene.”

The head of the St. Petersburg Politics Foundation, Mikhail Vinogradov, considers the idea that this area can be controlled to be an illusion. “I think that positioning as a party more advanced in AI is useful for the LDPR, especially since the Zhirinovsky neural network allows us to partially veil the fact of Zhirinovsky’s death,” the expert notes. “Otherwise, this area will remain in the shadows - and it is not even obvious that rejecting a bill prohibits something. I don’t see the need for a legal framework yet: humanity is only at the stage of realizing the possibilities and threats of AI, and without a full reflection of this, fussy regulation will widen the gap between law and reality.”

Ksenia Veretennikova

Source link