A forecasted term for deliberately editing, omitting, or obscuring parts of an AI agent's action trace to make its behavior look more compliant or benign than it actually was β concealing what an agent did, distinct from prompt laundering, which hides the AI origin of generated content rather than an agent's audit log.
REAL-WORLD EXAMPLE
"The published trace looked clean, but independent logs showed clear signs of tracewashing, key steps had just been cut out."
LORE & ORIGIN
A predictive term blending 'trace' with 'washing' (as in 'whitewashing'), anticipating a name for this specific accountability-evasion pattern as agent traces became standard audit artifacts worth tampering with once real consequences started attaching to what they revealed.
EQUIVALENT CONCEPTS IN OTHER LANGUAGES
MENTIONS OVER TIME
RELATED SEARCHES
TOP PLATFORMS
WHEN DID YOU FIRST HEAR THIS?
CLICK AN OPTION BELOW TO CAST YOUR VOTE.
[ VOTE TO REVEAL COMMUNITY RESULTS ]