Simulation of Language Evolution under Regulated Social Media Platforms: A Synergistic Approach of Large Language Models and Genetic Algorithms

2025-02-26

AI Summary

This study investigates how users in regulated social media platforms evolve evasive language strategies to circumvent content moderation. We propose the first LLM-driven multi-agent coevolutionary framework: a user agent optimizes evasion policies via genetic algorithms (GAs), while a regulatory agent concurrently enhances detection capabilities. Crucially, we decouple constraint enforcement (regulation) from expressive adaptation (evasion) and tightly integrate LLMs with GAs for policy search and refinement. Evaluated on two real-world scenarios—ciphertext-based communication games and illicit pet trading—we demonstrate significant improvements in information transmission accuracy and sustained dialogue length. A 40-participant user study confirms the ecological validity of the evolved strategies. Ablation studies reveal that the GA component is essential for long-term strategic adaptability, outperforming static or RL-based alternatives. Our work advances the understanding of adversarial language evolution under institutional constraints and provides a scalable, interpretable framework for studying human-AI coadaptation in moderated digital environments.

Technology Category

• Multiagent Systems: Adversarial Agents
• Machine Learning: Large Multimodal Models (LMMs)
• Game Theory and Economic Paradigms: Adversarial Learning

Application Category

• Economics, Online Markets and Human Computation: Uses of LLMs and GenAI for marketplace design, bidding, and strategic interactions
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Social media platforms frequently impose restrictive policies to moderate user content, prompting the emergence of creative evasion language strategies. This paper presents a multi-agent framework based on Large Language Models (LLMs) to simulate the iterative evolution of language strategies under regulatory constraints. In this framework, participant agents, as social media users, continuously evolve their language expression, while supervisory agents emulate platform-level regulation by assessing policy violations. To achieve a more faithful simulation, we employ a dual design of language strategies (constraint and expression) to differentiate conflicting goals and utilize an LLM-driven GA (Genetic Algorithm) for the selection, mutation, and crossover of language strategies. The framework is evaluated using two distinct scenarios: an abstract password game and a realistic simulated illegal pet trade scenario. Experimental results demonstrate that as the number of dialogue rounds increases, both the number of uninterrupted dialogue turns and the accuracy of information transmission improve significantly. Furthermore, a user study with 40 participants validates the real-world relevance of the generated dialogues and strategies. Moreover, ablation studies validate the importance of the GA, emphasizing its contribution to long-term adaptability and improved overall results.

Problem

• Simulate language evolution under social media constraints
• Develop multi-agent framework with LLMs and GAs
• Evaluate language strategies in regulated platforms

Innovation

• Large Language Models simulate evolution
• Genetic Algorithms optimize language strategies
• Multi-agent framework evaluates regulatory impacts