Prompt injection detection
Named by 4 kept items in this workspace.
The items that name it 4
injection-pattern scanning of AI agent instructions
Detects forged system turns, credential sniffing, and exfiltration instructions in model replies
defense: detect prompt injection attempts against agents
two-tier defense (heuristics + fine-tuned ML model) against injected instructions in fetched content
In every workspace you can read 5
All techniques in this workspace.