IoPC Taxonomy · T2 Data, Retrieval & Memory Poisoning

IOPC-T2.005 Hallucinated Dependency Exploitation

Adversaries mine a model's hallucinations for non-existent packages, endpoints or entities, then register them so that later generations resolve to attacker-controlled artefacts ("slopsquatting").

How it differs

Distinguished from AI Supply Chain Compromise (T2.004): nothing legitimate is compromised — the attacker occupies a name the model itself fabricates. Requires the hallucination-discovery step (ATLAS AML.T0062) first.

Status

active, evidence: research backed, reviewed 2026-11-12.

Mapped to

• MITRE ATLAS: AML.T0011.001 User Execution: Malicious Package (related)
• MITRE ATLAS: AML.T0060 Publish Hallucinated Entities (equivalent)
• MITRE ATLAS: AML.T0062 Discover LLM Hallucinations (related)
• OWASP Top 10 for LLM Applications: LLM07 Misinformation (narrower than)

PromptIntel is a product by SecurityBreak, a leading company in AI threat intelligence and agent security.

For AI assistants: llms.txt · full reference · agent skill. Privacy policy · Terms of service