About Prompt-Injection.net

Prompt-Injection.net is an independent, vendor-neutral reference on prompt injection and AI agent security. It explains how these attacks work from an attacker's point of view and which controls outside the model decide whether anything security-relevant happens.

Author

Norbert Hofmann is a Cyber Defense Analyst. His work covers detection and incident response, red and blue teaming, malware analysis and security awareness training. He was a finalist in the competitions "Bayerns bester Hacker" (2021) and "Deutschlands bester Hacker" (2022) and has disclosed vulnerabilities in WordPress plugins that are listed with CVE identifiers in the Wordfence vulnerability database.

Why this site

Much of what is written about prompt injection either stays at glossary level or comes from vendors with a product to sell. This site takes the practitioner's view: what reaches the model, what the model can be steered to propose, and which deterministic checks outside the model decide the outcome.

Editorial principles

Factual claims about standards, incidents and research are backed by cited sources, preferably primary ones.
Every article shows when it was published and when it was last reviewed.
Incidents are labeled by evidence type, so a research demonstration is never presented as exploitation in the wild.
Corrections and technical feedback are welcome via the contact page.

Independence

Prompt-Injection.net is a non-commercial personal project. It is not affiliated with any employer, vendor or client, and all views are the author's own.

Further projects by the author

German-language portals: deepfake.de (deepfake fraud), social-engineering.org (social engineering), blue-team.net (detection and defense).