A newly recognized AI assault method can let unauthenticated customers set off privileged workflows and entry enterprise methods, highlighting a niche in how identification and entry controls apply to AI brokers, in line with analysis from Noma Labs.
The report, authored by Noma Labs lead researcher Sasi Levi, describes the difficulty as “workflow identification hijacking,” the place attackers bypass customary controls by sending regular, benign requests via an unauthenticated entry level corresponding to a help inbox, GitHub difficulty, internet type, or shared doc.
“The enterprise AI pipeline reads the enter, interprets the request, and executes the motion precisely as designed,” Levi wrote within the report. “The core failure is that the requester had no authority to make that request.”
