What open-source LLM security tools does AISecIntel Group offer?

AISecIntel Group provides a modular suite of open-source security tools including PromptGuard (input validation & sanitization middleware), RAG Faithfulness Scorer (hallucination detection), Vector Isolation Auditor (tenant separation), and LLM Supply Chain Auditor (dependency and weights vulnerability scanner). Our repositories are hosted publicly at [https://github.com/AISecIntelGroup-Tools](https://github.com/AISecIntelGroup-Tools).

How does AISecIntel Group mitigate prompt injection vulnerabilities?

Our flagship tool, PromptGuard, uses a pluggable low-latency pipeline to sanitize unicode homoglyphs, strip invisible spoofing characters, calculate token entropy, and run heuristic scanners to catch direct overrides and indirect prompt injections (OWASP LLM01).

Does AISecIntel Group provide custom AI security consulting or audits?

Yes, we provide collaborative adversarial assessments, LLM penetration testing, and custom guardrail middleware integration for scaling project teams. Inquiries can be routed directly to our engineering team at kazemeyni@aisecintelgroup.com.

Where can I read about OWASP Top 10 vulnerabilities for LLMs?

The AISecIntel Knowledge Hub contains detailed, deep-dive publications covering the complete OWASP Top 10 for Large Language Models framework, including detailed defensive guides on Prompt Injection, Sensitive Info Disclosure, Model DoS, and System Prompt Leakage.

Sensitive information disclosure in LLMs and how to stop it

Dr. Fatemeh Kazemeyni

6/18/20261 min read

CONTACT

security@aisecintelgroup.com

@ 2026 AISecIntel Group.

SUBSCRIBE

AISecIntel Group
Open Source Adversarial AI Defense