Authentication based on detection of user-specific authentication input errors
In some embodiments, a computing system may monitor authentication input and modify authentication requirements based on detection of user-specific input errors. The computing system may use machine learning or other techniques to detect whether an incorrect authentication input corresponds to a common input mistake of a user. If the incorrect authentication input does correspond to a common input mistake of the user, a computing system may modify one or more authentication requirements to make the authentication process easier for the user.
1. A computer security and machine learning system for monitoring password entry and modifying login requirements based on detection of user-specific typing errors, the system comprising:
one or more processors programmed with computer program instructions that, when executed by the one or more processors, cause operations comprising:
detecting a password field has input focus of a user device associated with a user;
based on detecting the password field has input focus, monitoring the password field for input of the user;
determining, based on monitoring the password field, that the input is an incorrect password of the user;
inputting the incorrect password into a machine learning model that has been trained on a dataset comprising typing errors made by a plurality of other users and typing errors made by the user;
generating, via the machine learning model, output indicating whether the incorrect password is a common typing error of the user; and
based on the output indicating that the incorrect password is a common typing error of the user, modifying a login requirement associated with the user by temporarily increasing a number of failed attempts that are allowed for password entry for the user.
2. The system of claim 1 , wherein modifying the login requirement associated with the user further comprises temporarily removing a requirement to answer one or more security questions associated with the user.
3. The system of claim 1 , wherein the instructions, when executed cause operations further comprising:
based on the output indicating that the incorrect password is a common typing error of the user, disabling a requirement for two-factor authentication; and
based on receiving a correct password, approving a login of the user without requiring two-factor authentication.
4. The system of claim 1 , wherein the output comprises a probability score and modifying the login requirement associated with the user comprises:
determining that the probability score is less than a threshold score; and
based on determining that the probability score is less than a threshold score, modifying the login requirement associated with the user.
5. A method comprising:
determining, based on monitoring authentication input, an incorrect authentication input associated with a user;
inputting the incorrect authentication input into a machine learning model that has been trained on a dataset comprising user input errors associated with the user;
generating, via the machine learning model, output indicating whether the incorrect authentication input is a common user input error for the user; and
based on the output indicating that the incorrect authentication input is a common user input error for the user, modifying an authentication requirement associated with the user.
6. The method of claim 5 , wherein modifying the authentication requirement comprises temporarily increasing a number of failed attempts that are allowed for authentication entry for the user or disregarding the incorrect authentication input as a failed authentication attempt.
7. The method of claim 5 , wherein the output comprises a semantic matching score between the incorrect authentication input and a corresponding correct authentication input, and wherein determining that the incorrect authentication input is a typing mistake that corresponds to the user comprises:
based on determining that the semantic matching score satisfies a threshold score, determining that the incorrect authentication input is a typing mistake that corresponds to the user.
8. The method of claim 5 , wherein modifying the authentication requirement comprises temporarily removing a requirement to answer one or more security questions associated with the user.
9. The method of claim 5 , further comprising:
receiving second authentication input comprising a second incorrect authentication input of a user;
generating, based on the second incorrect authentication input and via the machine learning model, output indicating that the second incorrect authentication input is not a typing mistake that corresponds to the user; and
based on the output indicating that the second incorrect authentication input is not a typing mistake that corresponds to the user, sending a notification to a monitoring system.
10. The method of claim 5 , wherein modifying the authentication requirement associated with the user comprises disabling a requirement for multi-factor authentication.
11. The method of claim 10 , further comprising:
receiving input comprising a correct authentication input; and
based on receiving the correct authentication input and based on determining that the incorrect authentication input is a typing mistake that corresponds to the user, authenticating the user without requiring multi-factor authentication.
12. The method of claim 5 , wherein the output comprises a probability score and modifying the authentication requirement associated with the user comprises:
determining that the probability score is less than a threshold score; and
based on determining that the probability score is less than a threshold score, modifying the authentication requirement associated with the user.
13. A non-transitory, computer-readable medium comprising instructions that when executed by one or more processors, causes operations comprising:
determining, based on monitoring authentication input, an incorrect authentication input associated with a user;
inputting the incorrect authentication input into a machine learning model that has been trained on a dataset comprising user input errors associated with the user;
generating, via the machine learning model, output indicating whether the incorrect authentication input is a common user input error for the user; and
based on the output indicating that the incorrect authentication input is a common user input error for the user, modifying an authentication requirement associated with the user.
14. The medium of claim 13 , wherein modifying the authentication requirement comprises temporarily increasing a number of failed attempts that are allowed for authentication entry for the user or disregarding the incorrect authentication input as a failed authentication attempt.
15. The medium of claim 13 , wherein the output comprises a semantic matching score between the incorrect authentication input and a corresponding correct authentication input, and wherein determining that the incorrect authentication input is a typing mistake that corresponds to the user comprises:
based on determining that the semantic matching score satisfies a threshold score, determining that the incorrect authentication input is a typing mistake that corresponds to the user.
16. The medium of claim 13 , wherein modifying the authentication requirement comprises temporarily removing a requirement to answer one or more security questions associated with the user.
17. The medium of claim 13 , further comprising:
receiving second authentication input comprising a second incorrect authentication input of a user;
generating, based on the second incorrect authentication input and via the machine learning model, output indicating that the second incorrect authentication input is not a typing mistake that corresponds to the user; and
based on the output indicating that the second incorrect authentication input is not a typing mistake that corresponds to the user, sending a notification to a monitoring system.
18. The medium of claim 13 , wherein modifying the authentication requirement associated with the user comprises disabling a requirement for multi-factor authentication.
19. The medium of claim 18 , further comprising:
receiving input comprising a correct authentication input; and
based on receiving the correct authentication input and based on determining that the incorrect authentication input is a typing mistake that corresponds to the user, authenticating the user without requiring multi-factor authentication.
20. The medium of claim 13 , wherein the output comprises a probability score and modifying the authentication requirement associated with the user comprises:
determining that the probability score is less than a threshold score; and
based on determining that the probability score is less than a threshold score, modifying the authentication requirement associated with the user.