Stored Prompt Injection in Agentic Memory Systems
Attackers can poison agent memory to strike across future sessions without returning.
Priya Sundareswaran
Red Team Lead & Staff Writer
Priya built and ran offensive AI testing programs for two enterprise security firms before turning to full-time journalism. She brings a practitioner's rigor to her coverage of LLM red teaming methodology and the evolving tooling used to stress-test frontier models.
1 story
Attackers can poison agent memory to strike across future sessions without returning.