Prompt Injection as an Authorization Bypass Vector
LLM agents expose full credentials to attackers through hidden text.
Seo-Yeon Park
Contributing Editor
Seo-Yeon Park
Seo-Yeon Park researches autonomous and multi-agent software systems, with a particular focus on how classical authorization assumptions break down when principals are non-human. She previously led an applied research team studying AI pipeline security before moving into editorial work.
1 story