US 12,626,257 B2Granted

Intelligent technical protocol based approach leveraging AI-ML to block vishing scammers

AssigneeBank of America Corporation

InventorsSivashalini Sivajothi, Maneesh Kumar Sethia, Boddu Vikas Teja, Ankit Kumar Sahoo

Life of this patent

Estimating the term…

Loading patent…

Abstract

Systems and methods detect and prevent vishing attacks through an integrated framework combining SIP header customization, STIR/SHAKEN frameworks, AI/ML analysis, and real-time speech analysis using the Viterbi algorithm. The system begins with call initiation, embedding authentication information in the SIP header. The SIP data is transmitted and verified using STIR/SHAKEN frameworks, ensuring the authenticity of the caller's identity. Verified data is cross-referenced with third-party databases and analyzed by an AI/ML engine to detect anomalies. If potential fraud is detected, the call is blocked, and the customer is notified. Calls that pass initial checks are further analyzed using the Viterbi algorithm, which converts speech to text and identifies suspicious patterns. An anomaly pattern detector processes the converted text to detect vishing indicators, terminating the call if a match is found. This multi-layered approach ensures robust protection against vishing, enhancing the security and reliability of voice communications while safeguarding users from fraud.

Claims20

Dependent claims

An information-security method for detecting and preventing vishing attacks, comprising the steps of:

initiating a call and sending a Session Initiation Protocol (SIP) invite from an originating service provider, wherein the SIP invite includes detailed authentication information in a SIP header comprising caller identity, caller location, and call initiation timestamp;
transmitting call data using a Secure Telephone Identity Revisited and Signature-based Handling of Asserted Information Using toKENS (STIR/SHAKEN) framework, which employs digital certificates to verify authenticity of a calling number, ensuring that a caller ID has not been spoofed;
receiving SIP data at a destination service provider and verifying the call data that has been transmitted using a verification service, which decodes the SIP header containing a JSON web token (JWT) with attestation details, origin and destination identifiers, and timestamps, to confirm validity of the calling number;
cross-referencing the SIP data as verified with a Caller Name (CNAM) database to retrieve additional caller identification information;
storing verified SIP data and CNAM information in a third-party database for further analysis, ensuring that all authentication and verification details are available for subsequent processing;
analyzing the SIP data and CNAM information using an artificial intelligence and machine learning (AI/ML) engine to detect anomalies comprising calls from blocked regions, abnormal call durations, and new calls from unusual locations, by comparing the SIP data against known patterns of legitimate and fraudulent calls;
blocking the call and notifying a customer if the AI/ML engine detects potential fraud based on the analysis of the SIP data and CNAM information, preventing the fraudulent call from reaching the customer;
allowing the call to proceed if the AI/ML engine is indecisive and performing real-time speech analysis using a Viterbi algorithm, which converts spoken language into text by determining a most probable sequence of states given a sequence of observed acoustic features, including phonemes and words;
processing converted text along with the SIP data and CNAM information using an anomaly pattern detector to identify vishing patterns comprising sudden spikes in call volume, typical vishing patterns, differences in voice signal ranges, and requests for sensitive information, by analyzing both linguistic content and call metadata;
terminating the call and informing the customer if the anomaly pattern detector identifies a match with known vishing patterns, thereby preventing continuation of the call and alerting the customer to an attempted scam;
and
continuing to monitor and process the call if no suspicious patterns are detected, ensuring ongoing security by periodically re-evaluating the call data and updating the AI/ML engine with new information to adapt to evolving vishing tactics.

9 more elements

An information-security system for detecting and preventing vishing attacks, comprising:

an originating service provider configured to initiate a call and send a Session Initiation Protocol (SIP) invite, wherein the SIP invite includes detailed authentication information in a header comprising caller identity, caller location, and call initiation timestamp;
a Secure Telephone Identity Revisited and Signature-based Handling of Asserted Information Using toKENS (STIR/SHAKEN) framework integrated with the originating service provider, configured to transmit call data and employ digital certificates to verify authenticity of a calling number, ensuring that a caller ID has not been spoofed;
a destination service provider configured to receive SIP data and verify the call data that has been transmitted using a verification service, wherein the verification service decodes the SIP header containing a JSON web token (JWT) with attestation details, origin and destination identifiers, and timestamps to confirm validity of the calling number;
a Caller Name (CNAM) database configured to cross-reference the call data as verified and retrieve additional caller identification information;
a third-party database integrated with the destination service provider, configured to store verified SIP data and CNAM information for further analysis;
an artificial intelligence and machine learning (AI/ML) engine configured to analyze the SIP data and CNAM information to detect anomalies comprising calls from blocked regions, abnormal call durations, and new calls from unusual locations, by comparing the SIP data against known patterns of legitimate and fraudulent calls;
a call blocking module integrated with the AI/ML engine, configured to block the call and notify a customer if the AI/ML engine detects potential fraud based on the analysis of the SIP data and CNAM information;
a speech analysis module configured to allow the call to proceed if the AI/ML engine is indecisive, and perform real-time speech analysis using a Viterbi algorithm, which converts a spoken language into text by determining a most probable sequence of states given a sequence of observed acoustic features, including phonemes and words;
an anomaly pattern detector integrated with the speech analysis module, configured to process the text that was converted along with the SIP data and CNAM information to identify vishing patterns comprising sudden spikes in call volume, typical vishing patterns, differences in voice signal ranges, and requests for sensitive information, by analyzing both linguistic content and call metadata;
a call termination module configured to terminate the call and inform the customer if the anomaly pattern detector identifies a match with known vishing patterns, thereby preventing continuation of the call and alerting the customer to an attempted scam;
and
a continuous monitoring module configured to monitor and process the call if no suspicious patterns are detected, ensuring ongoing security by periodically re-evaluating the call data and updating the AI/ML engine with new information to adapt to evolving vishing tactics.

9 more elements

Dependencies are inferred from the claim text.