# US 12,602,501 B2 · System and method for generating synthetic data > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,602,501 B2 - Document ID: US12602501B2 - Status: Granted - Kind: B2 - Application number: 18/788,040 - Filed: 2024-07-29 - Published: 2026-01-29 - Granted: 2026-04-14 - Earliest claimed: 2024-07-29 - Assignees: Bank of America Corporation - Applicants: Bank of America Corporation - Inventors: Manu Kurian; Michael R. Young; Vinesh Premji Patel - Web page: https://patentgrep.com/patent/US12602501B2 ## CPC - G06F 21/6218: {to a system of files or objects, e.g. local or distributed file system or database} - G06F 21/53: by executing in a restricted environment, e.g. sandbox or secure virtual machine - G06F 2221/034: Test or assess a computer or a system - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) ## Abstract In response to receiving a request to generate synthetic data based on real-world data that is stored in a first memory, a processor accesses the first memory to extract at least a portion of the real-world data to use as sample data and determines data properties of the real-world data based on the sample data. The processor generates, based on the data properties, the requested synthetic data that at least partially mimics the real-world data. In response to detecting a request from an unauthorized user to access the real-world data, the processor provides the unauthorized user access to the synthetic data that mimics the real-world data. ## Claims ### Claim 1 A system comprising: a memory configured to store data properties of real-world data; and a processor communicatively coupled to the memory and configured to: receive a request to generate synthetic data based on real-world data stored in a first memory, wherein the synthetic data at least partially mimics the real-world data; in response to receiving the request, access the first memory to extract at least a portion of the real-world data, wherein the extracted portion of the real-world data is to be used as sample data for generating the synthetic data; determine data properties of the real-world data based on the sample data extracted from the first memory; generate, based on the data properties of the real-world data, the requested synthetic data that at least partially mimics the real-world data stored in the first memory, wherein the data properties associated with the synthetic data at least partially match the data properties associated with the real-world data; detect a request from an unauthorized user to access the real-world data in the first memory; and in response to detecting the request to access the real-world data, provide the unauthorized user access to the synthetic data that mimics the real-world data; wherein the processor is configured to generate the synthetic data by: obtaining a second piece of data, wherein the second piece of data comprises one or more of a second piece of real-world data or a second piece of synthetic data; inputting the real-world data and the second piece of data into a data tumbler, wherein the data tumbler is a software program configured to generate the synthetic data by mixing different pieces of data; and obtaining the synthetic data as an output of the data tumbler. ### Claim 2 (depends on 1) The system of claim 1, wherein the processor is further configured to: store the synthetic data in a second memory that is different from the first memory; and in response to detecting the request to access the real-world data, provide the unauthorized user access to the synthetic data stored in the second memory. ### Claim 3 (depends on 1) The system of claim 1, wherein: the real-world data is associated with a production system; the generated synthetic data is configured to at least partially mimic the real-world data associated with the production system; the request from the unauthorized user comprises a request to access the production system; and the processor is further configured to: generate a sandbox environment using the synthetic data, wherein the sandbox environment is separate from and configured to mimic the production system; and in response to detecting the request to access the real-world data, provide the unauthorized user access to the sandbox environment. ### Claim 4 (depends on 1) The system of claim 1, wherein production system runs on a first data server and the sandbox environment runs on a second data server that is different form the first data server. ### Claim 5 (depends on 1) The system of claim 1, wherein the processor is further configured to generate the synthetic data by obfuscating at least a portion of the real-world data. ### Claim 6 (depends on 1) The system of claim 1, wherein the processor is configured to use a generative Artificial Intelligence (AI) algorithm to generate the synthetic data based at least in part upon the real-world data. ### Claim 7 A method comprising: receiving a request to generate synthetic data based on real-world data stored in a first memory, wherein the synthetic data at least partially mimics the real-world data; in response to receiving the request, accessing the first memory to extract at least a portion of the real-world data, wherein the extracted portion of the real-world data is to be used as sample data for generating the synthetic data; determining data properties of the real-world data based on the sample data extracted from the first memory; generating, based on the data properties of the real-world data, the requested synthetic data that at least partially mimics the real-world data stored in the first memory, wherein the data properties associated with the synthetic data at least partially match the data properties associated with the real-world data; detecting a request from an unauthorized user to access the real-world data in the first memory; and in response to detecting the request to access the real-world data, providing the unauthorized user access to the synthetic data that mimics the real-world data; wherein the method further comprises generating the synthetic data by: obtaining a second piece of data, wherein the second piece of data comprises one or more of a second piece of real-world data or a second piece of synthetic data; inputting the real-world data and the second piece of data into a data tumbler, wherein the data tumbler is a software program configured to generate the synthetic data by mixing different pieces of data; and obtaining the synthetic data as an output of the data tumbler. ### Claim 8 (depends on 7) The method of claim 7, further comprising: storing the synthetic data in a second memory that is different from the first memory; and in response to detecting the request to access the real-world data, providing the unauthorized user access to the synthetic data stored in the second memory. ### Claim 9 (depends on 7) The method of claim 7, wherein: the real-world data is associated with a production system; the generated synthetic data is configured to at least partially mimic the real-world data associated with the production system; the request from the unauthorized user comprises a request to access the production system; and further comprising: generating a sandbox environment using the synthetic data, wherein the sandbox environment is separate from and configured to mimic the production system; and in response to detecting the request to access the real-world data, providing the unauthorized user access to the sandbox environment. ### Claim 10 (depends on 7) The method of claim 7, wherein production system runs on a first data server and the sandbox environment runs on a second data server that is different form the first data server. ### Claim 11 (depends on 7) The method of claim 7, further comprising generating the synthetic data by obfuscating at least a portion of the real-world data. ### Claim 12 (depends on 7) The method of claim 7, further comprising using a generative Artificial Intelligence (AI) algorithm to generate the synthetic data based at least in part upon the real-world data. ### Claim 13 A non-transitory computer-readable medium storing instructions that when executed by a processor cause the processor to: receive a request to generate synthetic data based on real-world data stored in a first memory, wherein the synthetic data at least partially mimics the real-world data; in response to receiving the request, access the first memory to extract at least a portion of the real-world data, wherein the extracted portion of the real-world data is to be used as sample data for generating the synthetic data; determine data properties of the real-world data based on the sample data extracted from the first memory; generate, based on the data properties of the real-world data, the requested synthetic data that at least partially mimics the real-world data stored in the first memory, wherein the data properties associated with the synthetic data at least partially match the data properties associated with the real-world data; detect a request from an unauthorized user to access the real-world data in the first memory; and in response to detecting the request to access the real-world data, provide the unauthorized user access to the synthetic data that mimics the real-world data; wherein generating the synthetic data comprises: obtaining a second piece of data, wherein the second piece of data comprises one or more of a second piece of real-world data or a second piece of synthetic data; inputting the real-world data and the second piece of data into a data tumbler, wherein the data tumbler is a software program configured to generate the synthetic data by mixing different pieces of data; and obtaining the synthetic data as an output of the data tumbler. ### Claim 14 (depends on 13) The non-transitory computer-readable medium of claim 13, wherein the instructions further cause the processor to: store the synthetic data in a second memory that is different from the first memory; and in response to detecting the request to access the real-world data, provide the unauthorized user access to the synthetic data stored in the second memory. ### Claim 15 (depends on 13) The non-transitory computer-readable medium of claim 13, wherein: the real-world data is associated with a production system; the generated synthetic data is configured to at least partially mimic the real-world data associated with the production system; the request from the unauthorized user comprises a request to access the production system; and wherein the instructions further cause the processor to: generate a sandbox environment using the synthetic data, wherein the sandbox environment is separate from and configured to mimic the production system; and in response to detecting the request to access the real-world data, provide the unauthorized user access to the sandbox environment. ### Claim 16 (depends on 13) The non-transitory computer-readable medium of claim 13, wherein production system runs on a first data server and the sandbox environment runs on a second data server that is different form the first data server. ### Claim 17 (depends on 13) The non-transitory computer-readable medium of claim 13, generating the synthetic data comprises obfuscating at least a portion of the real-world data.