The Weaponization of Generative AI in Disinformation Campaigns
Year: 2025
Type: article
Abstract: Generative AI reduces the price and speeds up disinformation, allowing multimodal creation and platform coordination in a brief time. This paper is an evaluation of its weaponization using a mixed-methods approach: (1) mass-detection of synthetic content and campaign form in the wild; (2) cross-modal detection and avoidance benchmark in text, image, audio, and video; (3) a preregistered experiment of belief, sharing, and confidence. In early windows, synthetic things spread out more easily, multi-modal composites more often defeat detectors, and exposure down-regulates the accuracy of belief and enhances sharing intent. Ensembles are also better detectors, although they perform poorly in realistic laundering (paraphrase, compression, re-recording, cross-modal remix). The network analysis indicates bridge accounts as the major cross-platform conduits. Our suggestions include provenance-by-default, calibrated detectors, coordination wasting, and human-in-the-loop validation of consequential claims. Results can be used in platform policy, election integrity protection, and multilingual risk monitoring in adaptive adversary efforts worldwide.
Source: Global Political Review
Author: Amrat Haq
Institution: International Islamic University, Islamabad
Language: English
Subfield: Sociology and Political Science
Field: Social Sciences
Domain: Social Sciences
Open Access status: hybrid
OpenAlex ID: W4415467693
FWCI:
Cites:
Cited by:
Related to: