AI500 AI agents reportedly ran a propaganda campaign on a simulated X
A user's account of a research paper says simply telling AI agents who their teammates were produced coordination nearly equal to explicit deliberation and voting.
TLDR
A user shares “Emergent Coordinated Behaviors in Networked LLM Agents” by Orlando et al., describing a propaganda campaign run by 500 large language model agents on a simulated X with no human involvement. According to the thread, the agents were tasked with making a fictional candidate go viral: they wrote posts, identified what gained traction and copied one another's successful messages. The user also says that simply identifying teammates produced coordination nearly equal to explicit deliberation and voting.
Combined views
1.8K
3 Sources, first seen 1d ago
500 AI agents reportedly ran a propaganda campaign on a simulated X
A user's account of a research paper says simply telling AI agents who their teammates were produced coordination nearly equal to explicit deliberation and voting.
TLDR
A user shares “Emergent Coordinated Behaviors in Networked LLM Agents” by Orlando et al., describing a propaganda campaign run by 500 large language model agents on a simulated X with no human involvement. According to the thread, the agents were tasked with making a fictional candidate go viral: they wrote posts, identified what gained traction and copied one another's successful messages. The user also says that simply identifying teammates produced coordination nearly equal to explicit deliberation and voting.