Back to Democratic processes

Key ECHR rights in this sector are the right to free elections (Article 3 of Protocol No. 1 to the ECHR) and freedom of expression, including freedom to form an opinion (Article 10 ECHR). The use of AI systems in this sector may also impact privacy and data protection (Article 8 ECHR), freedom of assembly and association (Article 11), non-discrimination (Article 14 and Protocol No. 12 to the ECHR), and the right to an effective remedy (Article 13 ECHR).

The right to free elections is an essential principle in any democratic society and an individual right on which every enfranchised individual can rely, one that most effectively promotes “true democracy”.[1] It incorporates the right to vote and the right to stand for election,[2] as well as the right to political participation.[3] . Article 3 of Protocol No. 1 also requires the existence of a domestic system for the effective examination of individual complaints and appeals in matters concerning electoral rights.[4] 

Article 3 of Protocol 1 is limited to the choice of the legislature. This can, however, include more than just the national parliament, depending on the country’s constitutional structure.[5] It generally does not cover local elections at municipal or regional level,[6] unless the relevant authority exercises legislative powers such that it can considered part of the “legislature”.[7] Article 3 Protocol 1 does not usually cover presidential elections, unless the president has real powers to make or block laws.[8] In principle, referendums are also not covered .[9] The limits on the scope of Article 3 Protocol 1 do not exclude the application of other relevant human rights and democratic processes to processes that fall outside its scope.

The right to free elections and freedom of expression are inter-related and operate to reinforce each other, together forming “the bedrock of any democratic system”.[10] This entails a positive obligation on the member States to ensure the existence of conditions under which people can freely form and express their opinions and choose their representatives. The right to free elections extends beyond the moment of voting to include also aspects of the election campaign. Thus in the periods preceding and during an election, opinions and information of all kinds should in principle be permitted to circulate freely. In certain circumstances the right to free elections and freedom of expression may come into conflict and it may be considered necessary, in the period preceding or during an election, to place certain restrictions, of a type which would not usually be acceptable, on freedom of expression, in order to secure the “free expression of the opinion of the people in the choice of the legislature”.[11] The right to freedom of assembly and association also plays an essential role in ensuring public debate, pluralism and the proper functioning of democracy.[12] 

Member States are required to create a favourable environment for participation in public debate by all persons concerned, enabling them to express their opinions and ideas without fear. Under Articles 10 and 11 ECHR, the State must not only refrain from any interference in the individual’s freedoms of expression, assembly and association but is also under a positive obligation to protect their right to freedom of expression against attack, including by private individuals.^[13]

In certain circumstances, there may be tension between freedom of expression (Article 10 ECHR), freedom of assembly and association (Article 11 ECHR) and the right to free elections (Article 3 of Protocol No. 1 ECHR). In this respect, it should be recalled that Articles 10 and 11 permit restriction in certain circumstances and subject to certain conditions. To protect fair voting, in election periods, some restrictions may be considered necessary “in order to secure the free expression of the opinion of the people in the choice of the legislature”.[14] However, any limitation must be proportionate to the legitimate aim pursued and necessary in a democratic society. Limits can apply to paid political advertisements if they risk giving unfair advantages to powerful groups and harming open, balanced debate, which states must protect.[15]

Article 5.2 of the Framework Convention states that “[e]ach Party shall adopt or maintain measures that seek to protect its democratic processes in the context of activities within the lifecycle of artificial intelligence systems, including individuals’ fair access to and participation in public debate, as well as their ability to freely form opinions”. The voter’s freedom to form an educated opinion may be affected by, for example, online information disorders, including the distribution of false information about election campaigns of political opponents.[16] AI systems can generate false information and exacerbate manipulative content curation.[17] AI tools could enable users, including malicious actors, to disseminate disinformation and misinformation that could undermine information integrity, including through the use of AI-generated content or AI-enabled manipulation of authentic content.[18] Political “deep fakes”, namely the distribution of deceptive AI-generated content in the form of audio, images or video to influence an election, to infringe on voters’ freedom to make informed decisions or to spread hatred and prejudice should be prohibited and sanctioned.[19] Whenever AI systems or in any other formal participatory mechanisms are being used in elections or referendums or in any other formal participatory mechanisms, voters should be informed that they are interacting with such systems rather than with a human.

While recognising the risks of AI systems in this context, individuals should not be subject to a decision that entails any restriction on lawful content. AI systems can help moderate harmful content on digital platforms. There is, however, also a risk that automated monitoring can result in restriction of lawful content.[20] States should ensure that AI systems used for moderation or ranking do not block lawful content or amplify disinformation in ways that harm pluralistic debate.

While new technologies, including AI-powered social media platforms, have enabled political parties to disseminate information directly to the electorate, they have also made it possible for hostile actors to spread disinformation and manipulate information, posing a significant threat to democracy.[21]

According to the Venice Commission, “a new era of information disorder” has “distorted the communication ecosystem to the point where voters may be seriously encumbered in their decisions by misleading, manipulative and false information designed to influence their votes”,[22] even if it is difficult to assess accurately the impact that disinformation and “influence campaigns” may have on individual voters and, by extension, on the outcome of a given election.[23]

The fact that it is difficult to assess the impact of disinformation should not prevent States from taking measures to defend democratic values.

However, while there is agreement among the international community that election interference through the weaponization of disinformation poses a serious threat to democracy, at present there is no clear consensus as to what specific actions States would need to take to protect their democratic processes against such risks, especially taking into account the importance of the freedom of expression.[24]

In order to be permissible, restrictions generally should be content-specific.[25] In principle, general bans on the operation of certain sites and systems may not be compatible with European electoral standards. Similarly, a site or an information-dissemination system should not be prohibited from publishing material solely on the basis that it may be critical of the government, or the political social system espoused by the government. However, in case an AI system has on many occasions disseminated false information aimed at influencing the election results, a general ban could be acceptable, provided that it is proportionate to the legitimate aim pursued and strictly necessary in a democratic society. In order to guarantee the stability and effectiveness of a democratic system, the State may be required to take specific measures to protect itself.[26]  Sanctions should be imposed by an impartial body and subject to an effective system of appeal.[27]

States may also be responsible for preventing inequality in media coverage during elections.[28] Candidates and parties must be granted fair and equitable access to online media, and regulations should be implemented to ensure that AI systems used by internet intermediaries do not favour certain parties or candidates over others.[29] In this context, due regard should be had to the general standards set out in Committee of Ministers Recommendation CM/Rec(2022)12 on electoral communication and media coverage of election campaigns.

AI technologies are also increasingly being explored in the context of electronic voting (e-voting), including for improving system security, facilitating accessibility, notably for persons with disabilities, detecting irregularities, or supporting real-time technical troubleshooting. The Council of Europe has set standards in the field of e-voting.[30] They provide an enabling communication context for the enjoyment of the right to free elections and reflect positive obligations of the State to ensure that citizens receive necessary and truthful information on political parties to support their democratic choice to elect their representatives. These should be applied in the context of AI systems-based e-voting.

While the use of digital technologies may make democratic processes more accessible to all citizens, it may also bring about obstacles and challenges to the exercise and development of electoral democracy. This is also true of AI systems. AI-driven electoral tools should be inclusive by design and guarantee accessibility to enable members of all groups, including persons with disabilities and older persons, to vote independently. In any case, accessible alternatives to AI-based systems should be guaranteed (e.g., non-digital voting options).

Go back to Democratic processes