BackSwarm Attacks

Swarm Attacks

Sub-category

Risk Domain

7AI System Safety, Failures & Limitations

Risks from multi-agent interactions, due to incentives (which can lead to conflict or collusion) and/or the structure of multi-agent systems, which can create cascading failures, selection pressures, new security vulnerabilities, and a lack of shared information and trust.

"Swarm Attacks. The need for multi-agent security is foreshadowed by attacks today that benefit from the use of many decentralised agents, such as distributed denial-of-service attacks (Cisco, 2023; Yoachimik & Pacheco, 2024). Such attacks exploit the massive collective resources of individual low- resourced actors, chained into an attack that breaks the assumptions of bandwidth constraints on a single well-resourced agent."(p. 39)

Entity— Who or what caused the harm

Human

Due to a decision or action made by humans

AI system

Due to a decision or action made by an AI system

Other

Due to some other reason or is ambiguous

Intent— Whether the harm was intentional or accidental

Intentional

Due to an expected outcome from pursuing a goal

Unintentional

Due to an unexpected outcome from pursuing a goal

Other

Without clearly specifying the intentionality

Timing— Whether the risk is pre- or post-deployment

Pre-deployment

Occurring before the AI is deployed

Post-deployment

Occurring after the AI model has been trained and deployed

Other

Without a clearly specified time of occurrence

Supporting Evidence (2)

"Such attacks are also used to great effect elsewhere, such as in ‘brigading’ on social media, in which teams of bots or humans collude to downvote or otherwise obstruct benign content (Andrews, 2021), or coordinated malicious actions in matching, rating, and content moderation"(p. 39)

"At present these bots are typically relatively unsophisticated but AI agents that can intelligently adapt and collaboratively identify new attack surfaces may amplify the potency of such attacks. More broadly, the ability for many small AI agents to parallelize tasks and recompose their outputs, such as in inference attacks that piece together sensitive information gathered individually by actors with limited access (Islam et al., 2012), can undermine the common assumption that individual agents with restricted capabilities are safe."(p. 39)

Part of Multi-Agent Security