A new chapter examines how generative AI is transforming scientific fraud and offers new insights into understanding and preventing these practices.
UNESCO has recently published a volume examining the challenges and transformations that artificial intelligence is bringing to scientific research. The volume includes Chapter 4, “Inteligencia artificial generativa y nuevas formas de engaño en la investigación científica”, co-authored by Cinta Gallent Torres, a member of the TeLL-ME research group, and Antoni Cerdà-Navarro, from the Universitat de les Illes Balears.
The chapter examines how generative artificial intelligence is changing not only the ways in which scientific knowledge is produced and communicated, but also the ways in which fraud and research misconduct can be committed.
Scientific fraud is not a new phenomenon, but its forms and scale have changed significantly. The authors trace its evolution from individual and largely manual forms of manipulation prior to 1990 to the progressive industrialisation of scientific fraud, including the emergence of paper mills, the sale of authorships, and the commercialisation of institutional affiliations.
The arrival of generative AI, particularly from 2022 onwards, introduces a new stage: what the authors refer to as “algorithmic fraud”. Artificial intelligence tools can facilitate certain fraudulent practices by reducing the time, costs, and technical knowledge required to generate or manipulate scientific content.
However, the problem is not exclusively technological. The chapter argues that understanding why such practices occur also requires attention to the factors that enable them and to the ways in which they may be rationalised.
The well-established “fraud triangle”, based on motivation, opportunity, and rationalisation, remains a useful framework for understanding research misconduct. Nevertheless, generative AI can expand both the opportunities to commit fraud and the ways in which such practices can be rationalised. Discourses such as “everyone does it” or “if I do not use these tools, I will fall behind” may contribute to normalising certain practices even before they take place.
For this reason, the authors propose extending the classical model by incorporating two additional dimensions: the capacity to carry out fraudulent behaviour and ethical awareness, understood as the ability to recognise its implications. AI-related scientific fraud can therefore be understood as a phenomenon shaped by technological, individual, institutional, and ethical factors.
The publication is particularly timely as universities, scientific journals, and research institutions are developing criteria for the responsible use of artificial intelligence. The challenge is not simply to determine which tools may be used, but also to establish how their integration into research can preserve the principles of integrity, responsibility, and scientific rigour.
The chapter is available in the UNESCO-published volume:
DOI: 10.54674/IPR.2026.97
