GRCLibrary.ai

Crosswalk

OWASP LLM Top 10 (2026)

Frameworks and control catalogues.

Where the corpus maps to OWASP LLM Top 10 (2026): Top 10 LLM application risks.

Canonical citation

Standard ID
OWASP LLM Top 10
Current version
2026
Publication date
2026-08
Topic
Top 10 LLM application risks
Upstream check location
https://genai.owasp.org/llm-top-10/ (opens in a new tab)
Last verified (UTC)
2026-08-30

Mapped control areas, 10

Control areas from the reverse crosswalk

From the Reverse Framework Control Crosswalk Matrix, which answers: given a control identifier or family, which documents address it.

Threat areaLibrary documentsAlignment type
LLM01:2026 Prompt injection (direct and indirect)AI Security and Risk Standard, AI and Agentic Development Security Standard, AI Adversarial Test Reference, AI Security Technical Implementation GuideArchitectural recommendation
LLM02:2026 Sensitive information disclosureAI Security and Risk Standard, Data Protection and Privacy Breach Response Procedure, Data Loss Prevention StandardArchitectural recommendation
LLM03:2026 Excessive agency in agentic systemsAI and Agentic Development Security Standard, AI Security Technical Implementation GuideArchitectural recommendation
LLM04:2026 Supply chain (model and component provenance)Third-Party AI Due Diligence Procedure, Software Composition Analysis StandardArchitectural recommendation
LLM05:2026 Data and model poisoningAI Security and Risk Standard, AI Model Risk Standard, AI Model Risk Assessment ProcedureArchitectural recommendation
LLM06:2026 Unbounded consumptionAI and Agentic Development Security Standard, AI Security Technical Implementation GuideArchitectural recommendation
LLM07:2026 Misinformation (including hallucination)Ethical AI Use Guideline, AI Evaluation Procedure, AI Testing, Validation and Documentation StandardArchitectural recommendation
LLM08:2026 Hidden context exposure (formerly system prompt leakage)AI Adversarial Test Reference, AI Security Technical Implementation GuideArchitectural recommendation
LLM09:2026 Vector and embedding weaknessesAI Security and Risk Standard, AI Security Technical Implementation GuideArchitectural recommendation
LLM10:2026 Improper output handlingAI Security Technical Implementation Guide, Developer Security RequirementsArchitectural recommendation

Coverage, 5 domains

Coverage by domain

From the Reverse Framework Control Crosswalk Matrix; the main crosswalk table's document counts land here.

Security, 1 document

Supply chain, 1 document

Privacy, 1 document

Dev security, 2 documents

AI, 9 documents

Corpus citations, 10

Documents naming it in their framework alignment

Generated from each document's framework alignment list in the taxonomy, through the committed alias map.

Note

What this mapping is

From the GRC Library Compliance Alignment Matrix: citations are indicative, not exhaustive (the representative-citation convention), and no entry constitutes a certification, attestation, or legal opinion.

From the Reverse Framework Control Crosswalk Matrix: coverage is partial, and external control identifiers and clause numbers are referenced for navigation only; the library does not reproduce third-party control text.

From the Cross-Framework Alignment Matrix: the alignment is not an official mapping, certification claim, legal opinion, or reproduction of third-party framework content.

The standards, laws, frameworks, and programmes in the canonical citations register, mapped here or not, are listed with their current editions on the Standards & frameworks page, generated from that register.