A method includes receiving marketing campaign associated data. The marketing campaign associated data is data associated with telephony communications from one or more telemarketing sources to a plurality of users. The method further includes applying a machine learning algorithm to the received marketing campaign associated data that determines anomalies associated with a subset of users of the plurality of users. The anomalies are determined based on audio signature of the telephony communications or based on telephony communications from a same source being disguised as different sources. The method includes identifying a marketing content within the telephony communications for the subset of users. The method also includes determining one or more common attributes between users within the subset of users and identifying a source of data breach based on the marketing content and further based on the one or more common attributes.