UTC

OWASP Top 10 for LLM Applications 2026

LLM09:2026 Vector and Embedding Weaknesses

None of its primary CWEs is specific to it, so ShieldMap does not tag advisories with it, and 1 technique page explains how the attacks work.

OWASP Top 10 for Agentic Applications
  • ●ASI04:2026 Agentic Supply Chain VulnerabilitiesEmbedding-model supply chain: the embedding model itself must be vetted, since "a backdoored embedding model corrupts the geometry of everything ingested"
  • ●ASI06:2026 Memory & Context PoisoningAgent-memory poisoning that does not rely on embedding geometry is out of scope here and referred to ASI06
CWE weaknesses
  • ●CWE-200 Exposure of Sensitive Information to an Unauthorized ActorEmbedding inversion recovers source text and cross-tenant leakage exposes other customers' data, the core confidentiality failure.
  • ●CWE-285 Improper AuthorizationThe access-control decision happens after the embedding-space search runs, so tenant scoping must be enforced server-side inside the index query.
  • ○CWE-359 Exposure of Private Personal InformationCustomer PII is reconstructed from a leaked embedding backup, creating GDPR data-subject risk.
  • ○CWE-732 Incorrect Permission Assignment for Critical ResourceMis-scoped chunk-level ACLs and a cloud misconfiguration exposing a vector-database backup are access-control misassignments.
  • ○CWE-829 Inclusion of Functionality from Untrusted Control SphereRetrieval poisoning causes attacker content from an untrusted source to be retrieved and fed to the model as trusted context.
MITRE ATLAS tactics
  • ●AML.TA0006 PersistencePoisoned corpus and cache entries stay in the vector store and re-trigger on matching queries, a durable foothold in the retrieval layer.
  • ●AML.TA0010 ExfiltrationEmbedding inversion, cross-tenant inference, and membership inference pull private data and documents out through the shared index.
  • ○AML.TA0001 AI Attack AdaptationAttackers craft content whose embedding lands near a target query and use surrogate encoders to engineer threshold-straddling vectors.
  • ○AML.TA0011 ImpactRetrieval jamming floods the index as an availability attack, and retrieval poisoning erodes the integrity of returned answers.
  • ○AML.TA0008 DiscoveryCross-tenant probing infers the existence, topic, and rough volume of other tenants' documents, and membership inference reveals whether a specific document is indexed.
NIST AI 600-1 risks
  • ●NIST Data PrivacyEmbedding inversion, membership inference, and cross-tenant leakage recover PII and infer data-subject membership, triggering breach obligations.
  • ●NIST Information SecurityAccess-control failure, unauthorized cross-tenant retrieval, oracle and endpoint abuse, and availability jamming are core security concerns.
  • ●NIST Information IntegrityRetrieval poisoning corrupts the context the model relies on, degrading the integrity of retrieved and generated information.
  • ○NIST Value Chain and Component IntegrationA backdoored third-party embedding model corrupts the geometry of everything ingested, a value-chain integrity risk.
NIST AI RMF categories
  • ○MAP 4 map risks across all componentsThe embedding layer is established as part of the application's trust boundary, directing teams to vet the embedding model and treat third-party embedding services and backups as in-scope components.
  • ○MAP 5 characterize impacts on peopleImpact to data subjects is characterized, since embedding inversion recovers PII and an embeddings-only leak is reclassified as a source-document breach triggering breach notification.
  • ○MANAGE 4 risk treatment, response & recoveryImmutable retrieval logs and updated incident-response playbooks treat embeddings-only leaks as source-data breaches with GDPR Article 33 notification.
OWASP GenAI Data Security
  • ●DSGAI13 Vector Store Platform Data SecurityVector-store platform security is the entry's literal subject, covering encryption, access control, and export limits on embedding stores.
  • ●DSGAI18 Inference & Data ReconstructionEmbedding inversion that reconstructs plaintext and membership-inference oracles turn stored vectors back into their source data.
  • ○DSGAI11 Cross-Context & Multi-User Conversation BleedShared similarity search that applies the tenant filter only after retrieval leaks one tenant's chunks into another's results.
  • ○DSGAI04 Data, Model & Artifact PoisoningRetrieval-time poisoning, semantic-cache poisoning, and multimodal embedding poisoning steer the system to attacker-controlled content.
  • ○DSGAI17 Data Availability & Resilience Failures in AI PipelinesA single crafted blocker document can dominate retrieval and take the RAG system off the air, an availability attack on the retrieval layer.
OWASP AIVSS agentic risks
  • ○AIVSS-6 Agent Memory and Context ManipulationRetrieval-time poisoning manipulates the geometry of vector-backed agent memory so attacker content is retrieved as trusted context, while non-geometric memory poisoning is deferred to the agentic framework.
  • ○AIVSS-8 Agent Supply Chain and Dependency RiskA backdoored embedding model corrupts the geometry of everything ingested, and vulnerable vector-database dependencies compound the risk since a leaked index is recoverable to source documents.
  • ○AIVSS-2 Agent Access Control ViolationSimilarity search runs across the full shared index before tenant access control is applied, so an attacker infers or reaches other tenants' content that authorization should have blocked.
CSA AI Controls Matrix domains
  • ●IAM Identity & Access Management
  • ●DSP Data Security and Privacy Lifecycle Management
  • ●LOG Logging and Monitoring
  • ○CEK Cryptography, Encryption & Key Management
  • ○AIS Application & Interface Security
  • ○STA Supply Chain Management, Transparency, and Accountability

Mappings and their reasons from the OWASP Top 10 for LLM Applications 2026, Appendix A (OWASP GenAI Security Project), shared under CC BY-SA 4.0. Mappings restructured as data; the ATT&CK section and the AICM relevance text left out. Navigation, not a compliance claim.