Similar principles apply to the question of whether messages generated by AI should enjoy speech or media rights. On the one hand, machines do not have any inherent rights to autonomy or self-expression because they lack an authentic self, and the potential for targeted communication to overwhelm an individual human’s information inputs is a new threat to human autonomy. But that is not to say that listeners or bystanders should not have the right to receive ideas disseminated by an automaton, or that the social value of democratic deliberation would not be served by nonhuman mediation. In many respects, however, there will be divergences. Data-driven AI permits the narrow and specific targeting of messages, which will give rise to a new ethics of propaganda: the targeting of limited and selected messages to individuals or narrow minorities raises the prospect of non-public mass communication. Contemporary concerns about fragmentation and filter bubbles are the tip of this normative iceberg. In time, a new ethics of communicative diversity and pluralism will be required to determine to what extent the privileges and duties of media should be enjoyed by AI.55
According to Harlem Desir, OSCE Special Representative on media freedom until 2020: ‘social media companies, search platforms and other internet intermediaries have responsibilities in dealing with problematic content, but they must not become the controllers of our fundamental human rights. We need to find the right balance in the demands to social media platforms . . . They are and should remain intermediaries, not publishers of every citizen’s opinions.’56
If media freedom matters – and the expanding universe of treaties, NGOs, IGOs and declarations focused on it suggests it does – then rules and codes which articulate and protect it must be based on an agreed definition of the media. Currently, they are not. The definition of ‘media’ will be essentially contested57 to the extent that media freedom engages special privileges and duties. A good deal of commentary – for example that of Kate Klonick58 – seeks to ask whether, on the basis of their current behaviour or performance, intermediaries such as internet platforms should be considered analogous to old media. An alternative approach would be to set out the privileges of media, and describe what intermediaries need to do in order to be considered media. This more ‘constructivist’ approach is adopted by the EU59 and the Council of Europe.60
Freedom From or Freedom To?
One of the key debates in freedom of speech theory is whether the freedom should be considered as a negative right, as freedom from state intervention, or whether it should be a positive right, a freedom to achieve certain communicative ends, which may entail some active intervention by the state to promote – even by intervening in the speech field – the conditions necessary to the enjoyment of this right.61 Such considerations also apply to media freedom. Many practical consequences depend on which approach is adopted: a negative rights approach prohibits intervention by state agencies into the speech field and would hold as incompatible with freedom of speech and media freedom any regulation of broadcasters which obliges them to express balanced or diverse views, or offer access to certain interests or perspectives.
As we will explore in subsequent chapters, divergence between European and international approaches – based on a more positive approach in Europe and more negative approaches in the US – can be traced to the founding texts. The First Amendment to the US Constitution obliges Congress to make no law to abridge freedom of speech or of the press (which has been held to include broadcasting). The freedom of expression guarantee in Article 10 of the ECHR, by contrast, expressly permits the licensing of broadcasting, and the ECtHR has held numerous times that there are positive obligations on the state to maintain the diversity and plurality necessary for democratic debate, including the maintenance of public broadcasting.62 It would be difficult to exaggerate the importance of this distinction.
State interference in the speech field can be presented as a restriction of media freedom, but the case can also be made that some form of subsidy or support of media is necessary for real enjoyment of communication and expression rights. Even in the US, a growing number of people are calling for new privileges and subsidies for journalism.63 A negative rights approach to social media would tend to focus attention on the need to find ways of limiting state incursions into the speech field, and ensuring that accountability and balancing of rights take place in courts in ways that prevent state control of media and public opinion. A positive rights approach could inform the opposite: namely that a regulatory framework might be applied to the media to ensure that they actively promote rights including the expression rights of their users. Because media both have expressive rights and are a necessary condition for the enjoyment of expressive rights of others, and because it is difficult to manage this dual role, it may be necessary to constrain them in the name of freedom of speech.
The question of what positive measures are needed to protect the media arises in particular in connection with journalist safety. Case law of the ECtHR has established that states have a positive obligation under Article 10 to ensure journalists’ safety and freedom from intimidation.64 Much of the policy debate about press regulation is a debate about accountability: how to institutionalize an accountability framework for the press without undermining autonomy. But the construction of structures for protection of journalists could have indirect effects, even if the journalist was not originally threatened by the state. When journalist Roberto Saviano, the author of Gomorrah, an exposé of organized crime in southern Italy, was threatened with Mafia reprisals, the home office minister stepped in to provide twenty-four-hour police protection. But when Saviano turned his fire on the corruption of that very minister, the Northern League’s populist rabble-rouser Matteo Salvini, veiled threats were made by the politician’s supporters that the minister could simply remove the armed guard. Saviano’s subsequent outspokenness required bravery. Experiments in Mexico and elsewhere have attempted to provide protection to journalists without compromising independence or chilling speech.
In a European context,65 the doctrine whereby states have a positive obligation to intervene to promote enjoyment of a right, rather than merely a negative obligation not to infringe a right, is increasingly applied to the media. According to the UN Principles on Business and Human Rights,66 such an approach should also be applied globally. The first cases are starting to be brought against social media platforms as censors of other people’s rights.67 The ongoing debate about positive versus negative approaches needs to be addressed directly, and the cul-de-sac of simplistic negative rights ‘content neutrality’ exposed.
Freedom from the State or from Private Actors?
Is media freedom protection from interference only by the state, or also from interference by private actors? Again, there is a distinction between the US approach and others around the world. US doctrine offers less protection from private interference with speech rights and stronger protection against state restrictions, whereas the European doctrines of positive interventions to protect freedoms, and the indirect horizontal effect of safeguards against interference with free speech by private actors, permit of more potential to protect people and media against private acts which have the effect of censoring speech, but may give states or courts stronger powers to decide what the limits on some people’s actions should be to protect the interests of others. The First Amendment to the US Constitution provides protection primarily against restriction of speech by law – which has been taken to mean the state more widely.68 Historically, because of First Amendment concern about any encroachment of state agencies in the speech field, public broadcasting has not benefitted from significant state-granted privileges or funding (aside from spectrum pricing).
Since the emergence of significant intermediaries with power to decide which messages are widely received, such as search and social media platforms in the first decade of the twenty-first century, states have been keen to delegate to those platforms various regulatory or censorship functions, with the result that there has been an expansion of private enforcement of censorship.69 From the point of view of