Deepfakes and bots in digital disinformation driven by malicious artificial intelligence

In an increasingly uncertain digital environment, how do we know if what we see is real? Deepfakes, which include videos with fake faces, synthetic audio, and manipulated documents, along with social bots that interact and persuade as if they were humans, represent a growing threat driven by malicious artificial intelligence. These technologies not only spread disinformation but also generate collective emotions such as fear, indignation, and distrust, affecting democratic stability and social cohesion. Faced with this reality, this article develops a bibliometric study and a scoping review of the literature (2014-2024) based on 316 documents indexed in Scopus, with the purpose of identifying trends, influential authors, methodological approaches, and research gaps on the use of artificial intelligence in disinformation. The results show an annual growth of 97%, an average age of 2.2 years per document, 19.7 average citations, and an h-index of 37, evidencing the sustained expansion and scientific interest in the detection of deepfakes and bots as central challenges of the digital ecosystem. Production is concentrated in India, the United States, and China. The co-occurrence analysis confirms the main research axes in deepfake detection, deep learning, and CNNs, along with the growing prominence of synthetic speech, diffusion models, and the economic and social effects of social media. Future research should address the role of AI in political communication, in particular its links to disinformation, polarization, and emotional influence, as well as the impact of the EU AI Act, the need for standardised metrics, and the emergence of transnational frameworks for digital governance.