The AI Security Wiki: A Free, Vendor-Neutral Framework for Securing AI
Most teams securing AI are rebuilding the same controls from scratch. I built the AI Security Wiki so you don't have to.
Most teams securing AI are rebuilding the same controls from scratch. I built the AI Security Wiki so you don’t have to.
It is a free, vendor-neutral framework for security leaders, architects and GRC professionals who are securing AI systems in real environments.
What is inside:
- 620 test cases for evaluating AI security products across 17 AI lifecycle layers and 6 emerging domains.
- 40 control objectives, each with evidence requirements and audit testing procedures.
- A 13-risk AI risk register with direct risk-to-control mapping for enterprise threat modeling.
- Domain-specific standards for Browser AI, Coding Assistants, Custom LLM Applications and Agentic AI.
- A research analysis of “AI Security” vs. “Security of AI”, and why the difference matters.
- A new GCC Regulatory Hub covering the UAE, Saudi Arabia, Qatar, Bahrain and Oman, with 4 role guides and 12 ready-to-use templates.
It is free to use and modify under CC BY-SA 4.0.
One request: the GCC regulatory content is compiled from available public sources and based on my current interactions in the region. If you work in the region, I would value your feedbacks, corrections if any and real-world validation.
Found a gap, a missing control or an edge case? Open an issue or suggest an edit from any page.
Explore the wiki: https://nachi065.github.io/AI-Security-Wiki/
What control or risk would you want to see added next? Tell me in the comments.