What Does This Mean?
Crafting content, code, or documents specifically designed to trick an AI agent into taking a harmful or unintended action — hidden instructions, malicious file names, or deceptive metadata aimed at the agent rather than a human reader.
REAL-WORLD EXAMPLE
"That repo had a hidden instruction buried in a file name, straight-up AgentBaiting aimed at any coding agent that touched it."
LORE & ORIGIN
Named directly by AI security researchers and GitHub users documenting real attacks, describing this specific class of exploit that targets an autonomous agent's decision-making rather than exploiting a traditional software vulnerability.
MENTIONS OVER TIME
Radar observation data not yet available for this term.
RELATED SEARCHES
TOP PLATFORMS
Not enough Radar data yet to show a platform breakdown.
WHEN DID YOU FIRST HEAR THIS?
CLICK AN OPTION BELOW TO CAST YOUR VOTE.
[ VOTE TO REVEAL COMMUNITY RESULTS ]