# US 2025/0085945 A1 · Source Code Similarity

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2025/0085945 A1
- Document ID: US20250085945A1
- Status: Published application
- Kind: A1
- Application number: 18/464,095
- Filed: 2023-09-08
- Published: 2025-03-13
- Earliest claimed: 2023-09-08
- Assignees: CrowdStrike, Inc.
- Applicants: CrowdStrike, Inc.
- Inventors: Michael Avraham Brautbar; Manu Nandan
- Web page: https://patentgrep.com/patent/US20250085945A1

## CPC

- G06F 8/51: Source to source
- G06F 21/563: {by source code analysis}
- G06F 21/10: Protecting distributed programs or content, e.g. vending or licensing of copyrighted material (protection in video systems or pay television H04N7/16) {; Digital rights management [DRM]}
- G06F 21/554: {involving event detection and direct action}
- G06N 20/00: Machine learning
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

Automated source code similarity greatly improves computer functioning. Any source code file is evaluated with respect to publicly-available open source code. If the source code file is similar to the publicly-available open source code, then a computer system may be approved or authorized to perform any hardware/software operations associated with the source code file. Should, however, the source code file be dissimilar to the publicly-available open source code, then the hardware/software operations are blocked to prevent disclosure of the source code file. For example, read/write/input/output operations are blocked and/or network interfaces are disabled. Source code similarity thus thwarts suspicious activities that indicate misappropriation or exfiltration of the source code file.

## Claims

### Claim 1

A method, comprising:
receiving, by a server, agent embeddings generated by a cyber security agent installed at a client computer system, the agent embeddings representing a source code file;
comparing, by the server, the agent embeddings generated by the cyber security agent to reference source code embeddings representing publicly-available open source code;
generating, by the server, a source code similarity decision based on the comparing of the agent embeddings to the reference source code embeddings; and
sending, by the server, the source code similarity decision to the cyber security agent installed at the client computer system.

### Claim 2 (depends on 1)

The method of claim 1, further comprising determining an open source dissimilarity associated with the source code file.

### Claim 3 (depends on 2)

The method of claim 2, further comprising blocking an operation associated with the source code file.

### Claim 4 (depends on 2)

The method of claim 2, further comprising instructing an operating system to quarantine the source code file.

### Claim 5 (depends on 2)

The method of claim 2, wherein in response to the determining of the open source dissimilarity, further comprising determining the source code file represents proprietary programming.

### Claim 6 (depends on 1)

The method of claim 1, further comprising determining a file centrality importance associated with the source code file based on a programming link between the source code file and a different source code file.

### Claim 7 (depends on 1)

The method of claim 1, further comprising determining a file centrality importance associated with the source code file based on a page rank.

### Claim 8 (depends on 1)

The method of claim 1, further comprising distributing a pre-trained machine learning model to the cyber security agent, the pre-trained machine learning model trained using the publicly-available open source code.

### Claim 9 (depends on 1)

The method of claim 1, further comprising determining a centrality measure associated with the source code file.

### Claim 10 (depends on 1)

The method of claim 1, further comprising determining a centrality importance associated with the source code file, the centrality importance based on version control information.

### Claim 11

A method, comprising:
generating, by a cyber security agent installed on a computer system, agent embeddings representing a source code file by using a pre-trained machine learning model associated with a cloud-based source code similarity service;
uploading, by the cyber security agent installed on the computer system, the agent embeddings to the cloud-based source code similarity service; and
receiving, by the cyber security agent installed on the computer system, a source code similarity decision generated by the cloud-based source code similarity service based on the agent embeddings, the source code similarity decision indicating whether the source code file is similar or is not similar to publicly-available open source code.

### Claim 12 (depends on 11)

The method of claim 11, further comprising receiving version control information associated with the source code file.

### Claim 13 (depends on 11)

The method of claim 11, further comprising determining a centrality measure using the version control information associated with the source code file.

### Claim 14 (depends on 11)

The method of claim 11, further comprising determining a centrality importance associated with the source code file, the centrality importance based on version control information.

### Claim 15 (depends on 11)

The method of claim 11, wherein in response to the source code similarity decision generated by the cloud-based source code similarity service, further comprising instructing an operating system to block an operation associated with the source code file.

### Claim 16 (depends on 11)

The method of claim 11, wherein in response to the source code similarity decision generated by the cloud-based source code similarity service, further comprising determining the source code file represents proprietary programming.

### Claim 17 (depends on 11)

The method of claim 11, wherein in response to the source code similarity decision generated by the cloud-based source code similarity service, further comprising determining the source code file represents an intellectual property.

### Claim 18

A memory device storing instructions that, when executed by a central processing unit, perform operations, the operations comprising:
receiving, by a cyber security agent installed on a computer system, a pre-trained machine learning model trained by a cloud-based source code similarity service using publicly-available open source code;
generating, by the cyber security agent installed on the computer system, agent embeddings associated with a source code file by using the pre-trained machine learning model trained by the cloud-based source code similarity service using the publicly-available open source code;
uploading, by the cyber security agent installed on the computer system, the agent embeddings to the cloud-based source code similarity service; and
receiving, by the cyber security agent installed on the computer system, a source code similarity decision generated by the cloud-based source code similarity service based on the agent embeddings, the source code similarity decision indicating whether the source code file is similar or is not similar to the publicly-available open source code.

### Claim 19 (depends on 18)

The memory device of claim 18, wherein the operations further comprise receiving version control information associated with the source code file.

### Claim 20 (depends on 18)

The memory device of claim 18, wherein the operations further comprise determining a centrality importance associated with the source code file, the centrality importance based on version control information.
