Granted Patent B1

US 12,513,160 · App. 18/145,470 · Granted Dec 30, 2025

Inventors: Norman Hoon Thian Poh (Atlanta, GA); Leanne Attard (Atlanta, GA)

Assignee: T Stamp Inc.

H04L63/1416H04L63/1425H04L63/1441

Abstract

Claims (62)

1 . A process comprising:

obtaining first data associated with at least one historical data attack;

encoding the first data into at least one fixed-size representation;

generating at least one anonymized vector representation based on the at least one fixed-size representation;

obtaining second data associated with a potential attacker;

encoding the second data into at least one additional fixed-size representation;

generating a second anonymized vector representation based on the at least one additional fixed-size representation; and

generating a confidence measure of the potential attacker as an attacker based on a comparison between the at least one anonymized vector representation and the second anonymized vector representation.

2 . The process of claim 1 , wherein:

the first data comprises a plurality of entries individually associated with a different historical data attack; and

the process further comprises:

generating a plurality of fixed-size representations based on each of the plurality of entries;

generating the at least one anonymized vector representation and a plurality of additional anonymized vector representations based on the plurality of fixed-size representations;

generating, via a clustering algorithm, at least one cluster based on the at least one anonymized vector representation and the plurality of additional anonymized vector representations; and

generating a confidence measure of the potential attacker as the attacker based on a comparison between the second anonymized vector representation and the at least one cluster.

3 . The process of claim 1 , wherein:

the first data comprises:

a first entry associated with a first data type; and

a second entry associated with a second data type;

the second data comprises at least one entry of the first data type and at least one entry of the second data type.

4 . The process of claim 3 , wherein:

generating the at least one anonymized vector representation comprises:

encoding the first entry into the at least one fixed-size representation;

encoding the second entry into a second fixed-size representation;

generating a second fixed-size representation based on the second entry; and

generating the anonymized vector representation based on the at least one fixed-size representation and the second additional fixed-size representation; and

generating the second anonymized vector representation comprises:

encoding the at least one entry of the first data type into the at least one additional fixed-size representation;

encoding the at least one entry of the second data type into a third fixed-size representation; and

generating the second anonymized vector representation based on the at least one additional fixed-size representation and the third fixed-size representation.

5 . The process of claim 3 , wherein the first data type comprises a media access control (MAC) address and the second data type comprises an internet protocol (IP) address.

6 . The process of claim 3 , wherein the first data type comprises a Subscriber Identity Module (SIM) number and the second data type comprises a device location.

7 . The process of claim 3 , wherein the first data type comprises a device operating system (OS) and the second data type comprises at least one of a MAC address, IP address, and Bluetooth address.

8 . The process of claim 3 , wherein the first data type comprises a device class and the second data type comprises a domain name system (DNS) request.

9 . The process of claim 3 , wherein:

the first data further comprises a third entry associated with a third data type; and

the second data further comprises at least one entry associated with the third data type.

10 . A system, comprising:

a server having a processor configured to:

receive, from a first computing device, first data associated with at least one historical data attack;

encode the first data into at least one fixed-size representation;

generate at least one anonymized vector representation based on the at least one fixed-size representation;

receive, from a second computing device, second data associated with a potential attacker;

encode the second data into at least one additional fixed-size representation;

generate a second anonymized vector representation based on the at least one additional fixed-size representation; and

generate a confidence measure of the potential attacker as an attacker based on a comparison between the at least one anonymized vector representation and the second anonymized vector representation.

11 . The system of claim 10 , wherein the processor is configured to:

receive a request associated with the potential attacker;

process the request to identify a target user account; and

transmit an alert to a service associated with the target user account.

12 . The system of claim 11 , wherein the alert causes the service to deny the request.

13 . The system of claim 11 , wherein the processor is configured to transmit an alert to a computing device associated with the target user account.

14 . The system of claim 11 , wherein the processor is configured to transition the target user account to a locked state.

15 . The system of claim 11 , wherein the processor is configured to enable a dual-authentication setting associated with the target user account.

16 . The system of claim 10 , further comprising at least one data store configured to store the first data, wherein the processor is configured to:

store the at least one anonymized vector representation at the at least one data store; and

based on the confidence measure of the potential attacker as the attacker:

store the second anonymized vector representation at the at least one data store; and

record, at the at least one data store, an association between the first data and the second data.

17 . The system of claim 16 , wherein, the at least one processor is configured to, based on the confidence measure of the potential attacker as the attacker:

generate a cluster based on the at least one anonymized vector representation and the second anonymized vector representation; and

store the cluster at the at least one data store.

Assignments (5)

Continuity (3)

References Cited (212)