| CPC G06F 21/6263 (2013.01) [H04L 63/1433 (2013.01)] | 20 Claims |

|
1. A system for obfuscating communications that include sensitive information based on context of the communications, the system comprising:
one or more memories; and
one or more processors, coupled to the one or more memories, configured to:
monitor an incoming communication, from a device and to a user device, associated with a user;
determine, using a machine learning model, a risk score associated with a context of the incoming communication, wherein the risk score indicates a likelihood that the incoming communication is associated with a fraudulent actor attempting to solicit sensitive information from the user, and wherein the risk score is based on:
a first content of the incoming communication,
a second content of an outgoing communication that is in response to the incoming communication, wherein the incoming communication and the outgoing communication include a voice call occurring in real-time, and
metadata associated with the incoming communication or the device;
monitor the outgoing communication, from the user device, for personally identifiable information (PII), wherein monitoring the outgoing communication includes performing one or more operations to secure the outgoing communication, and wherein the one or more operations are based on the risk score satisfying one or more thresholds;
detect, based on monitoring the outgoing communication and using natural language processing (NLP), a portion of the outgoing communication that includes the PII;
obfuscate the portion of the outgoing communication to generate an obfuscated outgoing communication by altering an audio characteristic of the voice call, wherein altering the audio characteristic comprises modifying an audio frequency of the portion of the outgoing communication to prevent unauthorized access to the PII within the outgoing communication; and
transmit, to the device, the obfuscated outgoing communication.
|