Skip to content

About

An open index of research on securing generative AI

The GenAI Security Literature Review is a community-driven, auto-updating database of research, standards, tools and resources on the security of large language models and AI systems. It currently tracks 1,386 resources across 46 categories in 8 domains.

How it works

  1. 01

    Auto-discovery

    A weekly GitHub Action collects new papers from arXiv, Semantic Scholar and CrossRef and proposes them in a pull request for human review before they are merged into the database.

  2. 02

    Curation

    Curated entries are categorized using the taxonomy, mapped to security frameworks (OWASP, NIST, MITRE ATLAS, ISO 42001) and tagged. Automatically added entries get keyword-based categories and stay marked as not yet reviewed until a maintainer vets them.

  3. 03

    Quality signals

    Entries marked "reviewed" have been vetted by a human; unreviewed entries carry a badge. Every change to the data is validated against the schema in CI. Citation counts, open-access status and venue help you prioritize what to read.

Coverage

Framework mappings

Entries are mapped, where applicable, to these frameworks:

OWASP Top 10 for LLM Applications
The most widely adopted LLM security risk list
OWASP Top 10 for Agentic Applications
Security risks specific to autonomous AI agents
MITRE ATLAS
Adversarial tactics, techniques and case studies for AI systems
NIST AI RMF
The US government's AI risk management framework
ISO/IEC 42001
International standard for AI management systems
Explore the mappings

Download the dataset

The full database is available for analysis, citation managers and systematic reviews. Files are regenerated on every deploy.

FormatAll entries (1,386)Reviewed only (95)
JSON complete recordsliterature.jsonfilter on reviewed
CSV spreadsheets, pandasliterature.csvliterature-reviewed.csv
BibTeX Zotero, LaTeXliterature.bibliterature-reviewed.bib

1,291 entries have reviewed: false: they were added automatically from academic APIs and have not been curated yet. Licensed under MIT; citation keys match the BibTeX shown on each entry page.

Contribute

Suggest a paper, tool or standard, or report a correction on GitHub.

Maintainer

Emmanuel Guilherme (@emmanuelgjr), OWASP contributor and GenAI data security researcher. Licensed under the MIT License.