Modified media detection

Patent metadata

Field	Value
Publication number	US-12086727-B2
Application number	US-202217978137-A
Country	US
Kind code	B2
Filing date	Oct 31, 2022
Priority date	Jun 29, 2020
Publication date	Sep 10, 2024
Grant date	Sep 10, 2024

Abstract

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for detecting modified media are disclosed. In one aspect, a method includes the actions of receiving an item of media content. The actions further include providing the item as an input to a model that is configured to determine whether the item likely includes audio of a user's voice that was not spoken by the user or likely includes video of the user that depicts actions of the user that were not performed by the user. The actions further include receiving, from the model, data indicating whether the item likely includes audio of the user's voice that was not spoken by the user or includes video of the user that depicts actions of the user that were not performed by the user. The actions further include determining whether the item likely includes deepfake content.

First claim

What is claimed is:

1. A computer-implemented method comprising:
accessing, by a computing device, historical data that includes data samples that each include a historical item of media content, a historical location of a historical computing device that received the historical item of media content, and a historical label indicating whether the historical item of media content includes deepfake content;
training, by the computing device, a model that is configured to receive a given item of media content and a given location of a given computing device that received the given item of media content and output data indicating whether the given item of media content includes deepfake content using machine learning and the data samples;
receiving, by the computing device, an item of media content, data identifying a location of a receiving computing device that received the item of media content, and data, from a user, indicating whether the item of media content includes deepfake content; and
updating, by the computing device, the model by retraining the model using machine learning, the historical data, the item of media content, the data identifying the location of the receiving computing device that received the item of media content, and the data indicating whether the item of media content includes deepfake content.

2. The method of claim 1 , comprising:
providing, by the computing device, the item of media content and the data identifying the location of the receiving computing device that received the item of media content as inputs to the model;
receiving, by the computing device and from the model, data indicating whether the item of media content likely includes deepfake content; and
providing, for output by the computing device and to the user, the item of media content and the data indicating whether the item of media content likely includes deepfake content,
wherein receiving the data, from the user, indicating whether the item of media content includes deepfake content is in response to providing, to the user, the item of media content and the data indicating whether the item of media content likely includes deepfake content.

3. The method of claim 1 , wherein:
the data samples each include data identifying a historical computing device that generated the historical item of media content, and
receiving the item of media content, the data identifying the location of the receiving computing device that received the item of media content, and the data indicating whether the item of media content includes deepfake content comprises receiving data identifying a generating computing device that generated the item of media content.

4. The method of claim 1 , wherein:
the data samples each include data identifying a historical computing device that received the historical item of media content, and
receiving the item of media content, the data identifying the location of the receiving computing device that received the item of media content, and the data indicating whether the item of media content includes deepfake content comprises receiving data identifying the receiving computing device.

5. The method of claim 1 , wherein:
the data samples each include data identifying a historical location of a historical computing device that generated the historical item of media content, and
receiving the item of media content, the data identifying the location of the receiving computing device that received the item of media content, and the data indicating whether the item of media content includes deepfake content comprises receiving data identifying a location of a generating computing device that generated the item of media content.

6. The method of claim 1 , wherein:
the data samples each include historical sensor data that reflects historical characteristics of a historical computing device that generated the historical item of media content, and
receiving the item of media content, the data identifying the location of the receiving computing device that received the item of media content, and the data indicating whether the item of media content includes deepfake content comprises receiving sensor data that reflects characteristics of a generating computing device that generated the item of media content.

7. The method of claim 1 , wherein:
the data samples each include historical sensor data that reflects historical characteristics of the historical computing device that received the historical item of media content, and
receiving the item of media content, the data identifying the location of the receiving computing device that received the item of media content, and the data indicating whether the item of media content includes deepfake content comprises receiving sensor data that reflects characteristics of the receiving computing device.

8. A system, comprising:
one or more processors; and
memory including a plurality of computer-executable components that are executable by the one or more processors to perform a plurality of actions, the plurality of acts comprising:
accessing historical data that includes data samples that each include a historical item of media content, a historical location of a historical computing device that received the historical item of media content, and a historical label indicating whether the historical item of media content includes deepfake content;
training a model that is configured to receive a given item of media content and a given location of a given computing device that received the given item of media content and output data indicating whether the given item of media content includes deepfake content using machine learning and the data samples;
receiving an item of media content, data identifying a location of a receiving computing device that received the item of media content, and data, from a user, indicating whether the item of media content includes deepfake content; and
updating the model by retraining the model using machine learning, the historical data, the item of media content, the data identifying the location of the receiving computing device that received the item of media content, and the data indicating whether the item of media content includes deepfake content.

9. The system of claim 8 , wherein the acts comprise:
providing the item of media content and the data identifying the location of the receiving computing device that received the item of media content as inputs to the model;
receiving, from the model, data indicating whether the item of media content likely includes deepfake content; and
providing, for output to the user, the item of media content and the data indicating whether the item of media content likely includes deepfake content,
wherein receiving the data, from the user, indicating whether the item of media content includes deepfake content is in response to providing, to the user, the item of media content and the data indicating whether the item of media content likely includes deepfake content.

10. The system of claim 8 , wherein:
the data samples each include data identifying a historical computing device that generated the historical item of media content, and
receiving the item of media content, the data identifying the location of the receiving computing device that received the item of media content, and the data indicating whether the item of media content includes deepfake content comprises receiving data identifying a generating computing device that generated the item of media content.

11. The system of claim 8 , wherein:
the data samples each include data identifying a historical computing device that received the historical item of media content, and
receivin

Assignees

• T Mobile Usa Inc

Inventors

• Yocam Eric
• Obaidi Ahmad Arash

Classifications

• G06F16/285

  Clustering or classification · CPC title

• G06N20/00

  Machine learning · CPC title

• G06N5/025

  Extracting rules from data · CPC title

• G06F21/64

  Protecting data integrity, e.g. using checksums, certificates or signatures · CPC title

• G06N5/04 Primary

  Inference or reasoning models · CPC title