MIT Technology Review AI highlights data lineage as a key security concern for companies deploying AI agents across business workflows.
In the scenario described by the publication, a sales representative asks an agent for customer pricing. The agent draws from several sources, including restricted pricing systems, internal communications and archived quotes. A second agent then uses the first agent’s output to draft a proposal, with no human checking either handoff.
The example shows how an answer that appears accurate to an agent can still carry forward data with different access restrictions, context or reliability. Tracking where information comes from—and how it changes as agents pass it between systems—can help organizations audit automated decisions and identify problematic data before it reaches a customer.