AgentDojo
Description: A benchmark environment for evaluating the security of language-model agents, particularly their vulnerability to prompt injection while using tools.
Primary topic: Adversarial Robustness in Machine Learning (55%)
OpenAlex ID: keywords/agentdojo
