This repository is a curated list of resources for AI security, including frameworks, attack methodologies, tools, and research papers. It serves as a comprehensive reference for researchers, security professionals, and developers interested in understanding and mitigating risks associated with artificial intelligence systems.

How It Works

The list categorizes resources into logical sections such as frameworks, taxonomies, offensive and defensive tools, adversarial techniques, privacy concerns, and research papers. This structure allows users to quickly navigate and find relevant information on specific AI security topics, from foundational standards to cutting-edge attack vectors and defense mechanisms.

Quick Start & Requirements

No installation or execution is required; this is a curated list of links and resources.

Highlighted Details

• Covers both offensive (e.g., adversarial attacks, prompt injection) and defensive (e.g., scanning tools, privacy-preserving libraries) aspects of AI security.
• Includes major industry standards and frameworks like NIST AI Risk Management Framework and ISO/IEC 42001.
• Features a broad range of tools for assessing ML model robustness, probing LLMs, and detecting AI-centric security risks.
• Compiles a significant collection of research papers on various AI security threats and countermeasures.

Maintenance & Community

The list is maintained by ottosulin, with an invitation for contributions via pull requests or direct contact. It is inspired by other "awesome" lists in the machine learning community.

Licensing & Compatibility

The repository itself is not software and does not have a license. Individual linked resources will have their own licenses.

Limitations & Caveats

As a curated list, the quality and maintenance of individual linked resources can vary. The rapidly evolving nature of AI security means some information may become outdated.