US 12,113,818 B2Granted

Machine learning for computer security

AssigneeCapital One Services, LLC

InventorsLawrence Douglas, Jeffrey Rule, Jackson Macomber

Life of this patent

Estimating the term…

Loading patent…

Abstract

A computing system may obtain text corresponding to a conversation between an outside caller and an agent. The computing system may obtain data associated with the conversation that may be used to determine whether the outside caller is attempting malicious activity or not. The obtained text and data may be provided to a machine learning model to generate a probability score indicative of whether the outside caller is attempting to obtain unauthorized access or attempting other malicious activity. Based on determining that the probability score satisfies a threshold, the computing system may modify a permission (e.g., an API permission) of the agent. The computing system may deny the request to perform the action and may cause display of an indication that the request was successful.

Claims18

Dependent claims

A system for improving cyber security for a software service by adjusting access permissions for agents of the software service based on detection of malicious activity, the system comprising:

one or more hardware processors programmed with computer program instructions that, when executed by the one or more hardware processors, cause operations comprising:
obtaining a data stream for a communication between a computing device associated with an agent of a service and an external device, wherein the data stream indicates use of a virtual private network by the external device;
processing the data stream using a machine learning model to generate a probability score indicative of whether the external device is attempting to obtain unauthorized access to the service;
based on determining that the probability score satisfies a threshold, modifying an application programming interface (API) permission of the agent, wherein the modifying removes an access permission of the computing device associated with the agent to perform an action;
based on modifying the API permission of the agent and based on receiving a request to perform the action from the computing device associated with the agent, denying the request;
and
based on denying the request, causing display, via the computing device associated with the agent, an indication that the request was successful.

4 more elements

A method comprising:

obtaining a data stream for a communication, wherein the data stream comprises natural language processing information of a conversation between a user and an agent;
obtaining data associated with the data stream, wherein the data indicates whether the user is attempting to obtain unauthorized access to a service associated with a computing device of the agent;
determining, based on the data and the natural language processing information, whether the user is attempting to obtain unauthorized access to the service,
wherein the determining comprises:
inputting the data and the natural language processing information into a machine learning model;
generating, via the machine learning model, an indication of whether the user is attempting to obtain unauthorized access;
and
based on the generated indication, determining that the user is attempting to obtain unauthorized access to the service;
based on determining that the user is attempting to obtain unauthorized access to the service, modifying a permission associated with the agent;
based on modifying the permission associated with the agent, denying a request to perform an action, wherein the request is received from a computing system associated with the agent;
and
based on denying the request, generating an indication that the request was successful.

9 more elements

A non-transitory, computer-readable medium comprising instructions that, when executed by one or more processors, causes operations comprising:

obtaining a data stream for a communication, wherein the data stream comprises natural language processing information of a conversation between a user and an agent;
obtaining data associated with the data stream, wherein the data indicates whether the user is attempting to obtain unauthorized access to a service associated with a computing device of the agent;
determining, based on the data and the natural language processing information, whether the user is attempting to obtain unauthorized access to the service,
wherein the determining comprises:
inputting the data and the natural language processing information into a machine learning model;
generating, via the machine learning model, an indication of whether the user is attempting to obtain unauthorized access;
and
based on the generated indication, determining that the user is attempting to obtain unauthorized access to the service;
based on determining that the user is attempting to obtain unauthorized access to the service, modifying a permission associated with the agent;
based on modifying the permission associated with the agent, denying a request to perform an action, wherein the request is received from a computing system associated with the agent;
and
based on denying the request, generating an indication that the request was successful.

9 more elements

Dependencies are inferred from the claim text.

Facts

Application: 17/812,428Filed 13 Jul 2022
Published: US 2024/0022582 A118 Jan 2024
Granted: 8 Oct 2024
Term ends: Not estimatedThe source records don’t support an estimate.

Classification4 codes

H04L 63/1425
By monitoring network traffic: traffic logging, e.g. anomaly detection
H04L 41/16
Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks: using machine learning…
H04L 63/107
For controlling access to devices or network resources: wherein the security policies are location-dependent, e.g. entities privileges depend on…
H04L
Digital communication

Family

US 2024/0022582 A1Pre-grant publication · published 18 Jan 2024
US 12,113,818 B2This patent

From the US filing history. Foreign family members aren’t listed.

Timeline

Citations

Open elsewhere

Building an agent or pulling many patents? Use the API or MCP server instead of scraping pages. Get an API key