Desinformação e inteligência artificial generativa: dinâmicas, atores e dilemas ético-informacionais em redes digitais

Abstract

The popularization since 2022 of generative artificial intelligence (AI) tools, such as ChatGPT, has amplified the disinformative potential of digital platforms by enabling the ultra-realistic creation of texts, images, audio, and videos of events that never occurred. While also used for recreational purposes, their appropriation by organized groups elevates the risks of public opinion manipulation, distortion of political narratives, and the weakening of democratic processes. In the Brazilian context, which is highly vulnerable to disinformation, the phenomenon demands an understanding of the interests, agents, and strategies that sustain its circulation. Grounded in the concept of the information regime, this research analyzes how actors, artifacts, devices, and actions structure the dissemination of manipulated content on digital platforms. The qualitative and descriptive investigation is based on the analysis of cases verified by Agência Lupa and on control mechanisms such as Proposed Law 2338/2023 and the policies on AI-generated content on the Google, Meta, TikTok, and X platforms. The results indicate a regime in which the economic logic of big techs and the fragility of regulatory mechanisms favor the activities of disinformation agents. The results shows that disinformation generated by generative AI is consolidating itself as an active force in the contemporary information regime, exploiting trust in public figures and being amplified by the economic logic of digital platforms.