# US 2026/0095474 A1 · Detecting bad server builds

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

- Number: US 2026/0095474 A1
- Document ID: US20260095474A1
- Status: Published application
- Kind: A1
- Application number: 18/901,453
- Filed: 2024-09-30
- Published: 2026-04-02
- Earliest claimed: 2024-09-30
- Assignees: Truist Bank
- Applicants: Truist Bank
- Inventors: Brian Matthew White; Ian Lassonde
- Web page: https://patentgrep.com/patent/US20260095474A1

## CPC

- H04L 63/1433: {Vulnerability analysis}
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)

## Abstract

A system and method for identifying bad server builds. The method scans the operating system software of a server while the server is in production, identifies an operating system software vulnerability, determines the publication date of the vulnerability, determines whether the publication date of the vulnerability is before the initial production date of the server and takes remedial action for future server builds if the publication date of the vulnerability is before the initial production date of the server.

## Claims

### Claim 1

A system for identifying bad server builds, said system comprising:
a back-end server including:
at least one processor for processing data and information;
a communications interface communicatively coupled to the at least one processor; and
a memory device storing data and executable code that, when executed, causes the at least one processor to:
train, using training test data, a neural network to predict factors for scanning operating system software on a server and identifying a vulnerability in the operating system software from the scan, the training including:
iteratively predicting the factors from the training test data that correlate to scanning operating system software on a server and identify a vulnerability in the operating system software from the scan, the predicting generating a prediction;
testing and comparing, during each iteration, the prediction to a target variable;
indicating, for each iteration and via a feedback loop, modifications to weights assigned to nodes of the neural network to improve the neural network's ability to predict the target variable and reduce error of the prediction;
deploy the trained neural network;
record a production date of when a server is first placed on a network and is in operation;
scan operating system software on the server while the server is in operation on the network using the deployed neural network;
identify a vulnerability in the operating system software from the scan using the deployed neural network;
determine if a publication date of the vulnerability is before the production date; and
take remedial action if the publication date of the vulnerability is before the production date to prevent future server builds from including the vulnerability before they are put into production.

### Claim 2 (depends on 1)

The system according to claim 1 wherein identifying a vulnerability and determining a publication date includes accessing a national vulnerability database.

### Claim 3

(canceled)

### Claim 4

(canceled)

### Claim 5 (depends on 1)

The system according to claim 1 wherein the server is part of a banking network.

### Claim 6

A method for identifying bad server builds, said method comprising:
training, using training test data, a neural network to predict factors for scanning operating system software on a server and identifying a vulnerability in the operating system software from the scan, the training including:
iteratively predicting the factors from the training test data that correlate to scanning operating system software on a server and identify a vulnerability in the operating system software from the scan, the predicting generating a prediction;
testing and comparing, during each iteration, the prediction to a target variable;
indicating, for each iteration and via a feedback loop, modifications to weights assigned to nodes of the neural network to improve the neural network's ability to predict the target variable and reduce error of the prediction;
deploying the trained neural network;
recording a production date of when a server is first placed on a network and is in operation;
scanning operating system software on the server while the server is in operation on the network using the deployed neural network;
identifying a vulnerability in the operating system software from the scan using the deployed neural network;
determining if a publication date of the vulnerability is before the production date; and
taking remedial action if the publication date of the vulnerability is before the production date to prevent future server builds from including the vulnerability before they are put into production.

### Claim 7 (depends on 6)

The method according to claim 6 wherein identifying a vulnerability and determining a publication date includes accessing a national vulnerability database.

### Claim 8

(canceled)

### Claim 9

(canceled)

### Claim 10 (depends on 6)

The method according to claim 6 wherein the server is part of a banking network.

### Claim 11

(canceled)

### Claim 12

(canceled)

### Claim 13

(canceled)

### Claim 14

(canceled)

### Claim 15

(canceled)
