Generative AI as a Disinformation Tool

The rapid spread of Large Language Models (LLMs) is transforming how individuals access and interpret information, increasing societal exposure to sophisticated forms of misinformation. As generative AI becomes a central information gatekeeper, it expands the attack surface for Foreign Information Manipulation and Interference (FIMI), enabling scalable data contamination, alignment manipulation, and realistic synthetic media. These dynamics weaken critical evaluation skills and amplify latent model biases, paralleling the long-term cognitive effects of traditional disinformation. Using lessons from science misinformation, the paper argues for robust regulation, transparent training practices, and integrated resilience strategies to mitigate emerging systemic risks posed by generative AI.