Fatimah Alhamzawi

Real-Time Detection and Mitigation of Prompt Injection Attacks in LLM-Integrated Enterprise Systems

Large language models (LLMs) embedded in enterprise workflows cannot structurally distinguish legitimate instructions from adversarial ones in the same token stream, making prompt injection OWASP's top LLM risk for two consecutive editions a persistent threat across direct and indirect vectors. This paper presents Prom...