This research introduces a multi-agent framework that simulates how language evolves when social media platforms enforce content moderation policies.

• Combines Large Language Models with genetic algorithms to model user adaptation strategies
• Creates realistic simulations of how users develop evasion techniques against platform regulations
• Demonstrates evolutionary patterns in language that emerge under continuous moderation pressure
• Provides insights for developing more effective, adaptive content moderation systems

Security Implications: This work helps platform security teams anticipate evolving evasion tactics, potentially improving detection systems while balancing expression freedoms and regulatory compliance.

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