# OWASP Top 10 for LLM Applications 2026

Framework id: `owasp_llm_top10_2026`. AI Security. Publisher: OWASP GenAI Security Project. Version: 2026 (August 2026). Region: Global. Mapping licence: CC BY-SA 4.0.
Source text: https://genai.owasp.org/resource/owasp-genai-llm-top-10-2026/

Each line is a clause, the NIST SP 800-53 controls OSA maps to it, and OSA's estimate of how far those controls cover it.
Clause titles are the source's own wording, from OWASP Top 10 for LLM Applications 2026, by the OWASP GenAI Security Project (August 2026), used unchanged under CC BY-SA 4.0. Coverage figures are OSA's own estimates.
How this mapping was made: OSA's own analysis, from OWASP's text by three independent ratings. NIST's catalogue of registered crosswalks has none from this list to SP 800-53 (checked 1 November 2026). The OWASP GenAI Security Project also keeps a crosswalk on GitHub that names SP 800-53 controls for each risk. Each of the ten risks was rated three times, independently, by AI models working from OWASP's text and the control statements. No rater saw another's rating. A control is listed for a risk where at least two of the three ratings named it. The coverage figure is the median of the three estimates of how much of the risk, and of OWASP's guidance for preventing it, the listed controls express. The rationale and gaps were written after the controls were settled, from the raters' reasons and from the gaps at least two of them named. Texts: OWASP Top 10 for LLM Applications 2026 (OWASP GenAI Security Project, Version 2026, August 2026).

## Clauses (10, average coverage 59%)
- LLM01 Prompt Injection: AC-03, AC-04, AC-06, AC-16, AC-24, AU-12, CA-08, CM-07, RA-03, SC-07, SI-10, SI-15 (55%)
- LLM02 Sensitive Information Disclosure: AC-03, AC-04, AC-23, AU-09, CA-08, IA-05, IR-08, IR-09, SC-04, SC-08, SC-28, SI-04, SI-12, SI-15, SI-19 (60%)
- LLM03 Excessive Agency: AC-03, AC-06, AC-24, AC-25, AU-12, CM-07, IA-02, SI-04, SI-10 (75%)
- LLM04 Supply Chain: CM-08, CM-10, CM-14, RA-05, SA-09, SA-10, SA-22, SI-02, SI-03, SI-07, SR-04, SR-06, SR-09, SR-11 (65%)
- LLM05 Data and Model Poisoning: AC-03, AC-04, AC-06, CA-08, CM-03, CM-14, SA-10, SC-07, SC-39, SI-04, SI-07, SI-10, SR-04, SR-06 (50%)
- LLM06 Unbounded Consumption: AC-12, AC-23, CM-06, IA-02, SC-05, SC-06, SC-07, SI-02, SI-04, SI-10 (60%)
- LLM07 Misinformation: AC-03, AC-06, AU-12, SA-11, SC-39, SI-10, SI-15 (35%)
- LLM08 Hidden Context Exposure: AC-03, AC-06, AC-25, IA-05, SC-03, SI-10, SI-15 (65%)
- LLM09 Vector and Embedding Weaknesses: AC-03, AC-16, AC-23, AU-09, AU-12, IR-08, SC-28, SC-32, SI-04, SI-10, SI-12 (55%)
- LLM10 Improper Output Handling: AC-04, AC-06, AU-12, CM-07, SA-11, SC-07, SC-18, SI-04, SI-10, SI-15 (70%)

## More
- A pattern's controls with their clauses in this framework: /api/v1/patterns/{id}/crosswalk?framework=owasp_llm_top10_2026
- Control-to-clause mappings as JSON: /api/v1/frameworks/owasp_llm_top10_2026?fields=mappings&per_page=100
- Rationale and gaps for each clause, as JSON: https://raw.githubusercontent.com/opensecurityarchitecture/osa-data/main/data/framework-coverage/owasp-llm-top-10-2026.json
- Page for people: /frameworks/owasp-llm-top-10-2026/
- Found an error? Open an issue at https://github.com/opensecurityarchitecture/osa-data/issues with the id, what OSA says and what the source says.
