Escalated authentication system to perform an integrity-check based on behavior patterns
Aspects of the disclosure include an escalated authentication system based on user behavior patterns. A user's behavior pattern on a device is collected and/or learned. The collected or learned pattern can be compared to subsequent behavior patterns to determine whether the current user is genuine or suspicious. Users deemed suspicious are subject to increased authentication requirements, often on-the-fly.
1. A method, comprising:
collecting telemetry data related to an interaction between a user and a device;
evaluating the collected telemetry data based on a standard dataset associated with a previous interaction between the user and the device;
determining whether the user is genuine or suspicious based on the evaluation;
enforcing a first authentication when the user is deemed genuine and enforcing a second authentication when the user is deemed suspicious;
adding the collected telemetry data to the standard dataset to create an updated standard dataset;
collecting additional telemetry data;
determining whether the user is genuine or suspicious based on evaluating the additional telemetry data on the updated standard dataset; and
increasing, based on the evaluation of the additional telemetry data, an authentication level on-the-fly when the user is determined suspicious.
2. The method of claim 1 , further comprising registering the user by collecting telemetry data for generating the standard dataset.
3. The method of claim 1 , wherein registering the user includes receiving an image from the user, generating a puzzle from the image, and presenting the puzzle to the user.
4. The method of claim 3 , wherein the telemetry data for the standard dataset includes interactions between the user and the device as the user solves the puzzle.
5. The method of claim 4 , wherein the interactions include one or more of:
which part of the puzzle was selected first;
which part of the puzzle was selected N-th;
which area of the puzzle is the user clicking;
is the user solving the puzzle from the middle or the corners;
time taken on moving one piece to a correct location; and
information related to scroll, drag and drop, and mouse click movements.
6. The method of claim 1 , further comprising determining that the user is suspicious when less than a threshold percentage of parameters from the collected telemetry data match the standard dataset.
7. The method of claim 1 , further comprising, based on how the collected telemetry data matches the standard dataset, imposing a single layer authentication, a dual-layer authentication, or a multi-factor authentication.
8. The method of claim 1 , further comprising performing machine learning on all collected telemetry data to predict to generate a machine model configured to predict whether the user is genuine or suspicious.
9. A non-transitory storage medium having stored therein instructions that are executable by one or more hardware processors to perform operations comprising:
collecting telemetry data related to an interaction between a user and a device;
evaluating the collected telemetry data based on a standard dataset associated with a previous interaction between the user and the device;
determining whether the user is genuine or suspicious based on the evaluation;
enforcing a first authentication when the user is deemed genuine and enforcing a second authentication when the user is deemed suspicious;
adding the collected telemetry data to the standard dataset to create an updated standard dataset;
collecting additional telemetry data;
determining whether the user is genuine or suspicious based on evaluating the additional telemetry data on the updated standard dataset; and
increasing, based on the evaluation of the additional telemetry data, an authentication level on-the-fly when the user is determined suspicious.
10. The non-transitory storage medium of claim 9 , further comprising registering the user by collecting telemetry data for the standard dataset.
11. The non-transitory storage medium of claim 9 , wherein registering the user includes receiving an image from the user, generating a puzzle from the image, and presenting the puzzle to the user.
12. The non-transitory storage medium of claim 11 , wherein the telemetry data for the standard dataset includes interactions between the user and the device as the user solves the puzzle.
13. The non-transitory storage medium of claim 12 , wherein the interactions include one or more of:
which part of the puzzle was selected first;
which part of the puzzle was selected N-th;
which area of the puzzle is the user clicking;
is the user solving the puzzle from the middle or the corners;
time taken on moving one piece to a correct location; and
information related to scroll, drag and drop, and mouse click movements.
14. The non-transitory storage medium of claim 9 , further comprising determining that the user is suspicious when less than a threshold percentage of parameters from the collected telemetry data match the standard dataset.
15. The non-transitory storage medium of claim 9 , further comprising, based on how the collected telemetry data matches the standard dataset, imposing a single layer authentication, a dual-layer authentication, or a multi-factor authentication.
16. The non-transitory storage medium of claim 9 , further comprising performing machine learning on all collected telemetry data to predict to generate a machine model configured to predict whether the user is genuine or suspicious.