The endless nature of social media feeds has led to a dramatic increase in the number of so-called “content farms”, where images and texts generated with Artificial Intelligence are created to achieve the greatest possible reach (see the infamous case of shrimp Jesus as an example). It is not uncommon to see perfectly smoothed faces with suspiciously homogeneous skin or strange body proportions, which at first glance may not even seem to indicate that something is wrong. In addition, it is also typical that strangely short, overly nice, massively appearing comments appear in the comments section, often as the activity of some automatic bot. These sites sometimes mix old, reliable posts with new, artificially generated content, sometimes even on hacked and renamed pages. Thus, users do not always realize that most of the posts they see in their news feed were essentially created from scratch (via AI tools).

Although these entries often seem harmless, the problem may increase in the future if Artificial Intelligence becomes even more advanced. It will then become more difficult for ordinary users to determine whether a photo, news item or even the person, artwork or event depicted in the image is real. At this point, the various “deepfake” technologies, which can manipulate not only images but also video and audio recordings in an astonishingly lifelike way, even by perfectly imitating the words and facial expressions of real people, can become particularly worrying. A speech attributed to a politician that was never actually made, or an alarming statement “borrowed” from a celebrity, can more easily deceive the public than ever before, increasing the risk of misinformation. The challenge for Meta is therefore how to maintain the credibility of the platform while increasing the risk of mass misinformation. The currently announced measure will likely have a negative impact on the spread of fake news, disinformation and misinformation, especially if Meta continues to move towards an aggressive and seemingly ill-conceived deregulation.

It is important to note that Facebook still has features that users find very valuable. Connecting in groups provides a more personal experience than many other social media platforms, the Marketplace allows for a broad and fast give-and-take, and many people find Facebook more credible than some older classifieds platforms on certain topics. Despite ongoing criticism and change, the company remains a major player, with billions of active users, guaranteeing the network’s reach and potential revenue.

The question is, of course, whether the company can find the optimal balance between control and user freedom. The challenge is illustrated by its efforts to minimize false deletions while avoiding a proliferation of harmful or grossly misleading content. Linked to this is the phenomenon of AI-generated posts, which could be a real concern if, in time, users have no means of distinguishing between real and fictional content. The changes Meta has now introduced will therefore not only increase freedom and community participation but will also pose challenges for the company and internet users alike. The coming months and years will determine whether Facebook will be able to keep pace with even more sophisticated AI solutions and how long it can maintain the platform as a relevant, real-connected community where the amount of artificial or manipulated information can be kept in check.

Meanwhile, in Europe, the Digital Markets Act (DMA) is significantly affecting Meta’s operations. According to the European Commission’s preliminary position, Meta’s “pay or consent” advertising model violates the DMA because it fails to offer users an alternative to receive the same quality of service with reduced personal data usage. Introduced in November 2023, the model requires users to either pay a subscription fee to avoid personalized ads or use the service for free while consenting to extensive data collection for ad personalization. The Commission argues that this approach does not comply with the principle of free user consent to data processing, as users are not provided with a true alternative that respects their privacy while maintaining service quality.

The current findings are preliminary, so Meta can defend itself and respond to the objections in writing. If the Commission confirms its presumption, it may find that Meta has infringed Article 5(2) of the DMA. Under the rules, a breach can result in a fine of up to 10 percent of world market turnover, and in the case of repeated breaches, this can rise to 20 percent. In addition, in extreme cases, a company could be required to divest certain businesses or be banned from buying additional related services if the rules are repeatedly ignored. The European Commission has said it intends to engage in a constructive dialogue with Meta to find a solution as soon as possible.

Meta appears to be moving towards a new, “softer” approach to content moderation and community in the United States, while regulators in Europe are setting stricter expectations regarding personal data and advertising models. Although the platform still has unique features and a large user base, the spread of AI-generated content and problems arising from data management regulations are posing serious dilemmas for the company. Meta’s current steps are therefore not only about how free its users can feel, but also about how well the company can comply with international regulations and changing technological challenges. The coming period will show whether it can develop a model that simultaneously protects the interests of users, complies with legal requirements, and makes the platform sustainable in the long term.

István ÜVEGES, PhD is a Computational Linguist researcher and developer at MONTANA Knowledge Management Ltd. and a researcher at the HUN-REN Centre for Social Sciences. His main interests include the social impacts of Artificial Intelligence (Machine Learning), the nature of Legal Language (legalese), the Plain Language Movement, and sentiment- and emotion analysis.