Machine learning-based method and system for detecting plaintext passwords
AssigneeSaudi Arabian Oil Company
InventorsAbdullah Abdulaziz Alturaifi, Nada Al-Noaimi, Sireen Abdullah Babateen
Life of this patent
Estimating the term…
Abstract
A technological solution for mitigating a cybersecurity risk on a computer that potentially includes a plaintext password. The solution includes searching a computer resource on the computer, analyzing any text, detecting a string of characters in the text that potentially includes a plaintext password, determining a confidence score for the string of characters indicating a likelihood the string of characters includes the plaintext password, and effectuating a remediation action based on the confidence score, wherein the remediation action includes encrypting the string of characters when the confidence score is equal to or greater than a first certainty level.
Claims20
A computer-implemented method for mitigating a cybersecurity risk on a target computer resource asset that potentially includes a plaintext password, the method comprising:
- identifying the target computer resource asset on a computer network;
- searching a computer resource on the identified computer resource asset;
- preprocessing any text in the computer resource using a plaintext character analyzer including a convolutional neural network (CNN) having a plurality of layers to generate a string of plaintext characters;
- analyzing the string of plaintext characters;
- determining a confidence score for the string of plaintext characters indicating a likelihood the string of plaintext characters includes a plaintext password;
- and
- effectuating a remediation action based on the confidence score,
- wherein the remediation action includes encrypting the string of plaintext characters when the confidence score is equal to or greater than a first certainty level.
A system having a machine learning platform for mitigating a cybersecurity risk on a target computer resource asset that includes a plaintext password, the system comprising:
- a processor and a memory, for implementing:
- a target computer resource asset identifying agent arranged to identify the target computer resource asset on a computer network;
- a plaintext search agent arranged to search text in a configuration file of a computer resource on the target computer resource asset;
- a plaintext character analyzer including a convolutional neural network (CNN) having a plurality of layers arranged to receive, preprocess, and analyze the in the configuration file to generate a string of plaintext characters;
- a plaintext password predictor arranged to detect a potential plaintext password in the string of plaintext characters and determine a confidence score that indicates a likelihood the string of plaintext characters includes a plaintext password;
- and
- a remediation unit arranged to effectuate a remediation action based on the confidence score,
- wherein the remediation action includes encrypting the string of plaintext characters when the confidence score is equal to or greater than a first certainty level.
A non-transitory computer readable storage medium containing cybersecurity risk mitigation program instructions for searching and analyzing computer resources in a target computer resource asset to detect and encrypt a plaintext password, the program instructions, when executed by a processor, causing the processor to perform an operation comprising:
- identifying the target computer resource asset on a computer network;
- searching a computer resource on the target computer resource asset;
- preprocessing any text in the computer resource using a plaintext character analyzer including a convolutional neural network (CNN) having a plurality of layers to generate a string of plaintext characters;
- analyzing any text in the computer resource;
- determining a confidence score for the string of plaintext characters indicating a likelihood the string of plaintext characters includes a plaintext password;
- and
- effectuating a remediation action based on the confidence score,
- wherein the remediation action includes encrypting the string of plaintext characters when the confidence score is equal to or greater than a first certainty level.
Dependencies are inferred from the claim text.