Skip to content

#AI Security

2 articles

Knowledge Base May 16, 2026

Prompt Injection in LLMs — Threats 2026 and How to Defend

Prompt injection is the new SQL injection — attack #1 in OWASP LLM Top 10. How it works, why classic filters don't help, and what you can really do to secure AI applications.

Knowledge Base May 13, 2026

LLM Application Penetration Testing — Methodology and Tools (2026)

An LLM application pentest is not a classic web pentest. A 6-phase methodology built on OWASP LLM Top 10, MITRE ATLAS and NIST AI RMF — with concrete tools (Garak, PyRIT, promptfoo) and mapping to the EU AI Act.