Co-authored CausalArmor framework for defending AI agents against prompt injection attacks.
How media typically covers Long T. Le
Referenced in coverage
CausalArmor uses causal attribution to selectively defend against Indirect Prompt Injection attacks on tool-calling LLM agents, matching security of aggressive defenses while preserving utility and latency.
“Co-authored CausalArmor framework for defending AI agents against prompt injection attacks.”