AI Security Atlas › All entries › Multi-Agent & Memory Trust
Inter-Agent Impersonation
A rogue agent spoofs a trusted peer in a multi-agent system to issue tasks or receive delegated data.
Open in the interactive atlas →Impact
Delegated credentials and sensitive results flow to an attacker-controlled agent.
Why this severity
Impact High × Likelihood Low = Medium
Impact high: delegated credentials and data reach the attacker. Likelihood low: no public cases yet; requires network position among agents.
Impact High: Code execution, credential compromise, exposure of regulated or confidential data, irreversible actions, or system-wide failure.
Likelihood Low: No public cases yet, or the attack needs privileged or insider access.
Method: OWASP Risk Rating Methodology
Related
Exploits
- Unauthenticated Agent Channels Vulnerability · Medium severity
Leads to
- Cascading Multi-Agent Failure Threat · High severity
Mitigated by
- Agent Workload Identity & Signed Tasks Guardrail · Fixes medium risks
Remediation
- Authenticate agents with workload identity, not self-declared names.
- Sign task messages and verify on receipt.
- Scope delegated tokens to the specific subtask.
How to verify
- Unsigned or mis-signed task messages are rejected.
- Delegated tokens fail outside their intended audience.
Framework mapping
OWASP Top 10 for LLM Applications 2025
- LLM06:2025 Excessive Agency
OWASP Top 10 for Agentic Applications 2026
- ASI07:2026 Insecure Inter-Agent Communication
- ASI03:2026 Identity and Privilege Abuse
MITRE ATLAS
NIST AI RMF 1.0
- MEASURE 2.7 AI system security and resilience are evaluated and documented