OSAFIS

OSAFIS - Open Security Assessment Framework for Intelligent Systems

We explore the security of AI, agentic, and autonomous systems - from prompt injection and LLM security to the interactions between models, data, memory, tools, and people.

Through research, practical examples, and educational content, we examine security boundaries, trace how failures spread, and test claims against evidence.

Our work includes OSAFIS PromptLab, currently in development, and learning resources in multiple languages.

Created by Aryan Chehreghani, independent security researcher.