Reinforcement-Learning Based Covert Social Influence Operations

Abstract

How might reinforcement-learning based covert social influence operations (CSIOs) be run, given that the CSIO agent wants to maximize influence and minimize discoverability of malicious accounts? And how successful can they be, given that both social platform bot detectors and humans might report them to the social platform? To answer these questions, we propose RL_CSIO, an RL-based methodology for running CSIOs and run 4 CSIOs with IRB-approval over a period of 5 days using a panel of 225 human subjects. We explore 8 research questions based on the data collected. The results show that RL_CSIO agents successfully trade off influence and discoverability - but in ways that are nuanced and unexpected.

BibTeX

@inproceedings{10.1145/3696410.3714729,
author = {Kumar, Saurabh and La Gatta, Valerio and Pugliese, Andrea and Pulver, Andrew and Subrahmanian, V.S. and Zhang, Jiazhi and Zhang, Youzhi},
title = {Reinforcement-Learning Based Covert Social Influence Operations},
year = {2025},
isbn = {9798400712746},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
url = {https://doi.org/10.1145/3696410.3714729},
doi = {10.1145/3696410.3714729},
abstract = {How might reinforcement-learning based covert social influence operations (CSIOs) be run, given that the CSIO agent wants to maximize influence and minimize discoverability of malicious accounts? And how successful can they be, given that both social platform bot detectors and humans might report them to the social platform? To answer these questions, we propose RL_CSIO, a methodology based on reinforcement learning (RL) for running CSIOs. We ran 4 CSIOs with IRB-approval over a period of 5 days using a panel of 225 human subjects. We explore 8 research questions based on the data collected. The results show that RL_CSIO agents successfully trade off influence and discoverability - but in ways that are nuanced and unexpected.},
booktitle = {Proceedings of the ACM on Web Conference 2025},
pages = {2435–2449},
numpages = {15},
keywords = {human-bot interaction, influence operations, social media},
location = {Sydney NSW, Australia},
series = {WWW '25}
}