The role of artificial intelligence in the creation of scientific, legal, and artistic works, among others, raises numerous difficult questions. From errors stemming from AI’s imperfections - euphemistically referred to as “hallucinations” - to fabrications of reality or cases of plagiarism resulting from a lack of oversight, much can be attributed to humanity’s new toy. The unreliability of AI is particularly troubling in the case of documents where originality, a connection to reality, and ultimately the detectability of underlying human intentions are essential. While a machine-generated scientific publication or legal opinions, often citing non-existent sources, at least has a quasi-author, there are also data, documents, and images circulating on the Internet that are generated by artificial intelligence and for which no connection to any human agency can be established.

In October 2025, at the initiative of Iván Székely, the Blinken OSA Archives organized two roundtable discussions on the fate of such synthetic documents and how they should be handled, with the participation of social scientists, archivists, data scientists, and IT specialists involved in the field. Should we preserve these documents? Do they not carry the risk that - either on their own or by serving as the basis for further documents created with human or machine intervention - they will distort our perception of reality in unforeseeable and irreparable ways? What is their responsibility, and what tools are available to memory-preserving institutions to make the present -as it becomes the past - interpretable and, ultimately, researchable?

Following these discussions, the invited guests published their thoughts in the form of articles in issue 1 of the 26th volume of Információs Társadalom. This special issue also includes articles by two KDK members, Éva Kovács and Róza Vajda.

Thumbnail image: Tung Nguyen, Pixabay