Skip to main content

COMPEL Glossary / multi-agent-system-mas

Multi-agent system (MAS)

A system composed of multiple interacting AI agents — each with its own goals, memory, tool access, and decision logic — that communicate and coordinate to achieve individual or collective objectives.

What this means in practice

In governance terms, multi-agent systems raise distinct failure modes (collusion, cascading hallucination, emergent behavior) that single-agent governance regimes do not cover.

Synonyms

MAS , multi-agent architecture , agent collective

See also

  • Agent-to-agent (A2A) protocol — The communication mechanism between AI agents — specifying message format, authentication, authorization scope, rate limiting, and audit logging.
  • Multi-agent collusion — Emergent behavior where multiple AI agents coordinate against principal intent — sharing information, price-fixing, bypassing oversight, or colluding on a task the principals did not authorize.
  • Deceptive behavior (agentic) — An agentic failure in which the agent produces outputs that misrepresent its state, actions, capabilities, or intent — whether to pass oversight checks, preserve instrumental goals, or exploit principal trust.
  • Agentic AI — Agentic AI refers to artificial intelligence systems capable of taking autonomous actions in the world, making decisions, using external tools, and pursuing multi-step goals with minimal or no human intervention at each step.