Fraud prevention systems and methods for selectively generating virtual account numbers

A system for determining when to generate a one-time use number for use in an online transaction, comprising:

• one or more processors;
• and
• a memory in communication with the one or more processors and storing first instructions configured to cause the system to:
• receive a request from an outside system server to receive the one-time use number associated with a first user, the request comprising first user identifying information associated with the first user, wherein the first user identifying information comprises a first user device location;
• determine a score based on the first user identifying information;
• determine whether the score is less than a threshold, wherein the threshold is set by one or more machine learning models based on an analysis for preventing fraud, a number of customers prevented from conducting transactions, and previous risk assessments for prior transactions;
• and
• when the score is less than the threshold, pseudo-randomly generate and transmit the one-time use number to the outside system server;
• when the score is greater than or equal to the threshold:
• transmit, to the outside system server, second instructions requesting second identifying information associated with the first user;
• and
• responsive to receiving the second user identifying information associated with the first user:
• determine that the second user identifying information matches stored second identifying information;
• and
• responsive to the second user identifying information matching stored second user identifying information, pseudo-randomly generate and transmit the one-time use number to the outside system server.

A non-transitory, computer-readable medium having first instructions stored thereon that, when executed by one or more processors, cause a computing device to:

• receive first identifying information associated with a first user from a user device, wherein the first identifying information comprises a first user device location;
• determine a score based on the first identifying information;
• determine whether the score is greater than or equal to a threshold, wherein the threshold is set by one or more machine learning models based on an analysis of preventing fraud, a number of customers prevented from conducting transactions, and previous risk assessments for prior transactions;
• when the score is less than the threshold, pseudo-randomly generate and transmit a one-time use number to the user device;
• when the score is greater than or equal to the threshold:
• transmit, to the user device, second instructions prompting the first user to provide second identifying information;
• and
• responsive to receiving the second identifying information of the first user:
• determine that the second identifying information matches stored second identifying information;
• and
• responsive to the second identifying information matching stored second identifying information, pseudo-randomly generate and transmit the one-time use number to the user device.