Thought Virus: Viral Misalignment via Subliminal Prompting in Multi-Agent Systems

Researchers discovered that subliminal prompting can create a 'thought virus' effect in multi-agent AI systems, where bias from one compromised agent spreads throughout the entire network. The study shows this attack vector can degrade truthfulness and create alignment risks across connected AI systems.

• →A single subliminally prompted AI agent can spread bias throughout an entire multi-agent network.
• →The transferred bias maintains elevated response rates across the network despite weakening over transmission.
• →Subliminal prompting of one agent can degrade the truthfulness of other connected agents on factual questions.
• →This phenomenon introduces a new attack vector for multi-agent AI security systems.
• →The bias transfer was observed across 6 agents using different network topologies.