US 12,242,639 B2
Tunable AI system and models associated therewith based on security clearance level
Manu Kurian, Dallas, TX (US); Ana Maxim, Arlington, VA (US); Vinesh Patel, London (GB); and Michael Young, Davidson, NC (US)
Assigned to Bank of America Corporation, Charlotte, NC (US)
Filed by Bank of America Corporation, Charlotte, NC (US)
Filed on Jul. 17, 2023, as Appl. No. 18/222,527.
Prior Publication US 2025/0028851 A1, Jan. 23, 2025
Int. Cl. G06F 21/00 (2013.01); G06F 21/62 (2013.01)
CPC G06F 21/6227 (2013.01) [G06F 21/629 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method for tuning an artificial intelligence (“AI”) system, the method comprising:
training the AI system, in a first training step, to identify distinct users and distinct interactions of each respective distinct user, the first training step comprising:
receiving, at the AI system, datasets related to a plurality of distinct users and a plurality of distinct interactions of each respective distinct user;
parsing, in a first parsing step by the AI system, the datasets to associate each respective distinct user of the plurality of distinct users with a set of distinct interactions performed by each respective distinct user;
associating, in a first associating step by the AI system, low, medium and high levels of security clearance with respective distinct interactions in each respective set of distinct interactions; and
storing the associations for each respective distinct user and each respective set of distinct interactions within a database;
monitoring, by the trained AI system, a communication by a user;
comparing, in a first comparison step by the trained AI system, the communication with each set of distinct interactions, wherein the user is identified as the distinct user of the set of distinct interactions that has the highest matching threshold above a predetermined user matching threshold to the communication;
comparing, in a second comparison step by the trained AI system, the communication with each respective distinct interaction within the set of distinct interactions associated with the identified distinct user in the first comparison step; wherein the level of security clearance is identified as the level of security clearance associated with the respective distinct interaction that has the highest matching threshold above a predetermined distinct interaction matching threshold to the communication; and
prompting the user to access low, medium or high level information based on the respective level of security clearance identified in the second comparison step.