Open Knowledge. Stronger Security.
AIHACKBins is an independent, open-source, community-driven knowledge base for hacking AI systems and hacking with AI. Our goal is to make AI security research accessible, practical, reproducible, and open to scrutiny.
Our philosophy draws inspiration from Kerckhoffs's principle: security should rely on resilient design and effective controls, not on keeping a system's workings obscure. Openness does not mean disclosing private data or credentials.
Research into the security of AI applications and their components, including prompt injection, jailbreaking, agents, MCP, retrieval-augmented generation, model APIs, multimodal inputs, and supply chains.
Reproducible ways to use AI in authorised security assessments: reconnaissance analysis, source-code review, API testing, vulnerability validation, security automation, and reporting.
- Free and open knowledge. No paywalls or exclusive research tiers.
- Practical research. Clear prerequisites, examples, observations, and limitations.
- Evidence first. Separate experimental claims from verified findings, including negative results.
- Attribution. Credit researchers and respect upstream licences.
- Collaboration. Invite questions, corrections, research, and review from anyone.
- Vendor neutrality. Evaluate techniques and tools on evidence, not sponsorship.
We are establishing the project and its contribution framework. You can participate by opening an issue, suggesting a topic, reviewing documentation, or proposing research through a pull request.
Project status: Founding phase. A public technique library and searchable website are planned but not yet released.
Knowledge improves through scrutiny.