Is there an active regulation to prevent deepfakes

Digital disinformation

Amélie P. Heldt

is a research assistant for a doctorate at the Leibniz Institute for Media Research, Hans Bredow Institute. She deals with freedom of expression on the Internet and researches the transformation of public communication.

The term "fake news" has been omnipresent for several years. In a narrower sense, this means deliberately false assertions of fact that can be spread with manipulative intent, especially before elections. In France, the legislature is trying to counter this.

View of the Assemblée nationale, the building of the National Assembly, the meeting place of the French Parliament. (& copy MAXPPP)

Fake news, disinformation, political microtargeting - these terms are often mentioned in the same breath as other Internet phenomena and confused in the public perception, even though they each concern different problems. Fake news are deliberately false statements of fact in the form of news that are distributed for political or financial reasons. According to this definition, fake news comes in three forms: messages that pay excessive attention to a topic, propaganda, and targeted disinformation. For example, half-truths that embellish a message or false statements of fact that are referred to as "alternative facts" are sufficient. Opinion manipulation rumors and lies existed long before the Internet. What is new about fake news is the massive and partly automated dissemination of fabricated news or facts that are at least not completely true or presented in misleading contexts. The network effect of the Internet works here: when people share content with their respective contacts and possibly across platforms, they give them a high distribution potential. [1] The infrastructure of the internet and the information intermediaries are used here in order to have a "viral" effect as quickly as possible. Disinformation campaigns and misinformation can be simplified or spread further using technical means, but this is not a sine qua non requirement for the work of those who try to influence the political discourse in a non-transparent way.

Fake news in the election campaign

The fear is that false news will be used deliberately during election campaign times and could have an impact on the election results because voters allow themselves to be manipulated by it. This danger is not entirely abstract, as it has already been observed in recent years that massive disinformation campaigns took place before the elections. Before the Brexit referendum in 2016, false facts and figures were circulating about the EU and Great Britain's membership, in the 2016 US presidential election campaign it was said that the candidate Hillary Clinton would direct child pornography, and in 2017 Emmanuel Macron was said to be homosexual and leading a fictitious marriage. In Germany, fear of fake news was great ahead of the 2017 federal elections, but empirical studies have shown that it has not materialized. In 2018 it was observed in Brazil that the massive sharing of false information in WhatsApp groups was an attempt to discredit the opposition candidate. Who is interfering? Although studies show that attempts to exert influence through fake news before elections tend to come from the right to the extreme right spectrum, no rule can be derived from this. There are indications that some of them come from abroad with the aim of interfering in the internal affairs of other countries in order to influence politics to their advantage or simply to destabilize the situation. The talk is above all of actions in relation to the US elections in 2016 from Russia (for example by so-called "troll armies") or in relation to the shift to the right in the EU (for example by Donald Trump's ex-advisor Steve Bannon). The motive for generating fake news can, however, also be of a financial nature, because it is mostly scandal headlines that are clicked on more quickly (so-called clickbaiting).

Digital services are used to try to influence people and automation can simplify and potentiate this process. But the actual effect of automation on election results, whether real or false, cannot be measured. One can use circumstantial evidence to speculate what concrete influence fake news has on elections, but the voting decision is made by people themselves and the reasons for this are not apparent from the outside. This applies to information in general, which is why it is so important for a democracy that citizens have access to reliable sources of information and that journalistic media serve the process of individual and public opinion-forming. However, when it comes to the problem of disinformation, the legislature can also take action.

What can the legislature do?

What is the solution to fake news if there is no clear, technical cause? The problem can be approached on several levels, for example with more media literacy or offers such as fact checks. [2] From the legislature's point of view, it is difficult to take regulatory action without disproportionately restricting fundamental rights such as freedom of expression or freedom of the press, precisely because the phenomenon is so diffuse and complex. France nevertheless decided to take this step: After false information about Emmanuel Macron was spread before the presidential elections in 2017, the French president announced a law against "fake news".

France already had two laws against undue electoral influence. 27 of the French Freedom of the Press Act of 1881 forbids "the publication, distribution or reproduction of false news, fabricated, falsified or untruthfully ascribed items to disturb the public peace for dishonest reasons or to create disturbance potential". So this is about fake news that is distributed with the intention of influencing public discourse. In the context of elections, Art. 97 of the French electoral law, with a prison sentence of ¼ up to a year or a fine of up to 15,000 euros for those who "use false news, defamatory rumors or other fraudulent maneuvers to confuse or falsify election results or discourage one or more voters from voting ".

The new regulation in France

The law against information manipulation was passed by the French parliament in November 2018 and in December 2018 the Conseil Constitutionnel (equivalent to the German Federal Constitutional Court) upheld it by rejecting pending complaints against the law. The draft law has been controversial since the French government's first draft and was only passed by parliament after a few changes. At first it was called (translated) "Law against False Information". In this version the prerequisite for an intention was missing, that is, it did not matter whether one knew about the untruth of the information and wanted to consciously disseminate it in order to influence elections. The lack of a subjective element has provoked a lot of criticism.

The final draft, the law against information manipulation ("loi contre les manipulations de l’information"), is directed less against false information per se than against the dissemination of false information for the purpose of electoral manipulation. The scope of application is defined in Art. 1: It is prohibited to make any assertion or inaccurate or misleading attribution about a fact that could affect the veracity of the upcoming elections and that is intentionally, artificially or automatically disseminated on a large scale via a public online communication service. " The law thus targets fake news that could reduce the credibility of the election results. The application of the law is limited in time to the three months before an election, including the day of the election. If a false message is discovered that corresponds to the definition in Art. 1, urgent proceedings can be initiated in front of the court. The regional court decides on the consequences of a violation within 48 hours upon request. The court can adopt all proportionate and necessary measures to prevent the spread of false information, i.e. blocking, deleting and / or not redistributing the content.

This means that the decision-making authority against individual actors remains essentially with the judiciary and the ban is not implemented by an authority. In the light of the separation of powers, this is welcome and ensures more trust in the rule of law. In contrast to the German Network Enforcement Act, the French legislature did not transfer responsibility to the social networks, which could otherwise exert a greater influence on public discourse and the election campaign.

Problems with this legislation

Although the French legislature has limited the potential to intervene in the law in terms of the scope and period of application, there could be collisions in relation to media freedoms. In journalism, for example, source protection (protected as a human right) applies: How can this be guaranteed when new transparency obligations are enforced, especially with regard to critical reporting? The so-called "scissors in the head" should also not be left unmentioned. Will citizens in the future be afraid of expressing their critical opinion on election campaign issues and therefore censor themselves? These aspects are relevant for both individuals and society, as freedom of opinion, information and media are closely interwoven in the process of forming opinions.

Apart from the questions that arise from a legal perspective, it is also unclear how practicable the law is in terms of implementation. As described, there is not just one form of fake news and it can therefore be difficult to unequivocally identify as such.This is all the more true when courts have little time and do not have the necessary technical knowledge. [3] The greatest challenge are "deepfakes", manipulated images or videos that look deceptively real and are created with the help of artificial intelligence. Even experts find it difficult to expose such deceptions. Even if speed is the norm in urgent proceedings, a court decision made on the basis of prima facie evidence, although the line between truth and falsification is unclear, could have far-reaching consequences. The tight time frame and the campaign situation could do their part. Regulations such as the French law against information manipulation during election campaign times may deter spreaders of fake news, but it cannot be determined whether they are really being deterred or whether their activities have a direct influence on the election results. [4] For this reason, cooperation with the platforms should not be neglected, because they can do their part. For example, Whatsapp can prevent messages from being forwarded en masse and Facebook can mark political advertisements more clearly. In this way, standards can arise from the architecture of the platforms, ie from the technology that is available, [5] which counteract the lack of transparency and manipulation of public discourse and contribute to ensuring a free and fair election process.