19 May 2024
Introduction:
Freedom of expression is part of the foundation of a democratic society. It applies not only to information and ideas that are favourably received or considered harmless, but also to those that shock, disturb or upset the state or particular segments of society. Such are the requirements of pluralism and broadmindedness, without which there is no democratic society.
At the same time, overstepping the permissible limits of freedom of expression, in the form of hate speech, disinformation, or manipulation of public opinion, are a threat to democratic values and peace. We are seeing worrying trends of hate speech, disinformation and manipulation of public opinion in virtual settings, as well as unacceptable measures against them taken by governments and companies that violate international human rights norms. In response to these trends, policies guaranteeing freedom of expression in virtual settings and curbing its abuse need to be reviewed and updated. This process should be based on the recommendations set out below, which summarise laws, policies and good practices adopted or in the process of being adopted in the European Union.
Recommendations:
1. Human rights that people have outside virtual settings should be protected in virtual settings too. This applies especially to civil liberties, including the right to privacy, the right to freedom of expression, religion, association, etc.
2. Freedom of expression in a virtual setting is the same as outside such a setting.
3. Freedom of expression in a virtual setting, like that in any other setting, is subject to such formalities, conditions, restrictions or penalties as are prescribed by law and are necessary in a democratic society in the interests of national security, territorial integrity or public safety, to prevent disorder or crime, to protect health or morals, to protect the reputation or rights of others, to prevent the disclosure of confidential information or to maintain the authority and impartiality of the judiciary.
4. The state has a duty to respect, protect and promote the realisation of human rights, which implies adopting policies and regulations that ensure freedom of expression in virtual settings and its legitimate restriction. All organizations, including business companies, and all citizens have an obligation to respect human rights in virtual settings.
5. In the design, adoption and development of policies and regulatory mechanisms, the state must ensure access to the drafting of such documents, transparency and effective citizen participation, through an open and inclusive process of consultation and/or public deliberation.
6. While hate speech per se is not defined or prohibited in international human rights law, it is illegal under certain conditions: when it incites discrimination, hostility or violence based on a protected characteristic. The state must establish legal means to hold such incitement accountable. Hate speech that does not reach the threshold of incitement, while not prohibited per se, can also be harmful and should be countered through appropriate educational measures.
7. The state should counter the spread of disinformation when it violates human rights through appropriate policies, including capacity building for prevention and resilience against forms of disinformation and raising media literacy of all citizens.
8. In using the term ” disinformation”, the state should strive to be very clear and avoid broad interpretations. The generally accepted definition of disinformation as demonstrably false information that is disseminated deliberately with the intent to cause serious social or individual harm should be seen as a boundary within which content can be specified but which should not be crossed. Unintentional errors, satires, parodies do not constitute disinformation. Clearly identified politically biased or propagandistic news and commentary is not disinformation per se. Content may not be defined as disinformation merely because it is false, offensive or harmful; or merely because it may provoke public opinion; or merely because it may harm the public calm or the national economy, affect the reputation of public institutions, public figures and other participants in public life; or merely because it constitutes rumours, myths or ‘conspiracy theories’.
9. Hostile and untruthful content, including disinformation, is not in itself illegal. They are included within the scope of protected expression and can only be restricted and sanctioned if they fall within the legitimate restrictions set out in the law. A blanket prohibition on the expression of untruths, erroneous opinions or misinterpretations of past events is not permissible.
10. Measures against the dissemination of disinformation must be consistent with international human rights law and the principles of legality, necessity and proportionality and must emphasize the importance of a free, independent, pluralistic and diverse media and of providing and promoting access to independent, factual and evidence-based information.
11. The state should discourage public officials from spreading hate speech or disinformation, and to this end should establish criteria in professional codes of conduct.
12. Any restriction that is deemed justified on the grounds that it incites discrimination, hostility or violence on the basis of a protected characteristic, or that is disinformation, must also be consistent with the general principles of limiting freedom of expression set out in Principle 3 above.
13. In restricting freedom of expression on the grounds of incitement to discrimination, hostility or violence on the basis of a protected characteristic, a specific assessment of the necessity and proportionality of the restrictive measure must be made by analysing the following factors: the context of the statement, the existence of a public debate on the subject, the status of the speaker, the status of the persons to whom the expression is directed, the intent to harm a vulnerable group, the content and form of the expression, the extent of its dissemination, and the likelihood of harm.
14. In adopting and implementing policies and regulations ensuring freedom of expression in virtual settings and its legitimate limitations, the state must ensure the protection of personal data and privacy in accordance with applicable law.
15. The state must ensure compliance with the prohibition of targeted advertising in a virtual setting based on profiling using sensitive data within the meaning of Article 9 of the GDPR: data concerning racial or ethnic origin; political opinions, religious or philosophical beliefs; trade union membership; genetic data; biometric data for the purpose of uniquely identifying an individual; data concerning an individual’s health, sex life or sexual orientation.
16. The state should take measures to ensure transparency of political advertising in the virtual space, including requiring the indication of the fact that the content is political advertising, as well as its authorship.
17. The state should establish mechanisms for independent oversight of technology companies aimed at protecting human rights and legal protections for users of these companies, with strict compliance with freedom of expression. In establishing such mechanisms, the state should respect legitimate commercial confidentiality requirements, and differentiate measures depending on the size and structure of companies.
18. All public and private creators of artificial intelligence have an obligation to ensure that any artificial intelligence clearly identifies itself as such prior to any interaction with a user that contains advertising of a product or may influence the user’s individual behaviour in the area of political participation.
19. Technology companies that provide digital services should:
a. Take due diligence measures to identify illegal content and activities on their platforms, and to counter hate speech, illegal disinformation and manipulative behaviour in virtual settings.
b. In the case of illegal content, implement policies and tools to moderate content, including measures such as removal, labeling, account suspension, content demonetization and deamplification, clickbait sanctioning, fact-checking initiatives in partnership with trusted flaggers, etc. These measures should meet the requirements of necessity and proportionality.
c. Organize a reporting mechanism allowing any natural or legal person to report certain information that they consider to be illegal content.
d. Provide priority consideration for notifications submitted by entities referred to as “trusted flaggers” who work in specific areas in which they have expertise.
e. Adopt and implement policies to counter illegal behavior in virtual settings, e.g., fake accounts, information amplification through bots, impersonation, troll farms, malicious deepfakes, etc.
f. Ensure transparency of their policies and procedures and provide detailed information on how restrictive measures are taken in different geographical areas and in different languages.
g. Provide periodic reports to a designated regulatory authority on the actions taken to address illegal content and conduct; these should include, for example, information on the number of orders they have been sent by government bodies and organisations in relation to content moderation or user information they have requested; a summary of the human resources involved in content moderation; the number of accounts and units of content that have been taken down by the company; the accuracy and error rate of their automated content moderation systems.
h. Ensure independent monitoring and measurement of the impact of restrictive measures.
i. Provide researchers with access to relevant data and information to enable them to understand the dynamics associated with hate speech and disinformation and to help inform evidence-based policy and regulatory measures.
j. Maintain independent mechanisms for handling complaints, including procedures for appealing or reviewing content removal, and providing consumers with effective remedies.
k. Continually introduce and improve features that give users greater control in a virtual setting, including dispute resolution mechanisms, transparency regarding content moderation decisions, and opportunities for users to challenge platform decisions.
l. Conduct due diligence regarding the effects of their products, policies and operations, in accordance with their responsibilities under the UN Guiding Principles on Business and Human Rights.
m. Companies that are very large online platforms (VLOPs) or very large online search engines (VLOSEs) within the meaning of Regulation (EU) 2022/2065 of the European Parliament and of the Council of 19 October 2022 (Digital Services Act) have the additional obligation to assess and take measures to address systemic risks arising from the design, operation and use of their services, as well as from the potential misuse of services by service recipients; in assessing systemic risks, they should focus on systems or other elements that may contribute to the risk, including any algorithmic systems that may be relevant, in particular their recommendation and advertising systems.
n. Systemic risks include, but are not limited to, (a) the risk associated with actual or foreseeable negative effects on democratic processes, civil discourse or electoral processes, and public safety; (b) the risk associated with actual or foreseeable negative effects on the protection of public health, minors, and serious negative effects on a person’s physical and mental well-being, or gender-based violence. Such risks may also arise from coordinated disinformation campaigns related to public health, or from online interface design that may encourage behavioural addiction of service recipients.
