Preventing ai engineered password compromises originating from keystroke rhythms
One example method includes automatically determining, at a keyboard sound mitigation model implemented on a first computing system, if the first computing system is actively running a video communication service during a video communication session with a second computing system; in response to determining that the first computing system is actively running the video communication service during the video communication session with a second computing system, automatically determining, by the keyboard sound mitigation model, if a user of the first computing system is entering a password or sensitive information into a webpage or window that is active on the first computing system; and in response to determining that the user of the first computing system is entering the password or the sensitive information, performing, by the keyboard sound mitigation model, a keyboard sound mitigation operation that masks or removes any keyboard sounds made while entering the password or sensitive information.
1 . A method, comprising:
automatically determining, at a keyboard sound mitigation agent that is implemented on a first computing system, if the first computing system is actively running a video communication service during a video communication session with a second computing system;
automatically determining, by a video communication monitor of the keyboard sound mitigation agent, whether a microphone of the first computing system is actively enabled, and enabling a password input monitoring function of the keyboard sound mitigation agent only when both (i) the video communication service is actively running and (ii) the microphone is actively enabled;
in response to determining that the first computing system is actively running the video communication service during the video communication session with a second computing system and that the password input monitoring function is enabled, automatically determining, by the keyboard sound mitigation agent, that a user of the first computing system is entering a password or sensitive information into a webpage or window that is active on the first computing system, wherein determining that the user is entering the password or sensitive information comprises (i) scanning the active webpage or window to detect a password input field or sensitive information input field, (ii) detecting that a cursor or user input focus has entered the password input field or sensitive information input field, and (iii) detecting entry of an initial keyboard event in the password input field or sensitive information input field; and
in response to determining that the user of the first computing system is entering the password or the sensitive information, performing, by the keyboard sound mitigation agent, a keyboard sound mitigation operation that masks or removes any keyboard sounds made while entering the password or the sensitive information, wherein the keyboard sound mitigation operation is initiated upon detection of the initial keyboard event and is terminated upon detecting that the cursor or user input focus exits the password input field or sensitive information input field, and wherein the keyboard sound mitigation operation is performed only while the microphone remains actively enabled.
2 . The method of claim 1 , further comprising:
in response to determining that the first computing system is actively running the video communication service during the video communication session with the second computing system, and before determining that the user of the first computing system is entering the password or the sensitive information, automatically determining, by the keyboard sound mitigation agent, if a microphone of the first computing system is actively turned on, and enabling or disabling a password input monitoring function of the keyboard sound mitigation agent based on whether the microphone is actively turned on, wherein disabling the password input monitoring function prevents scanning of the active webpage or window and monitoring of keyboard events.
3 . The method of claim 1 , wherein determining if the first computing system is actively running the video communication service during the video communication session with the second computing system comprises:
iterating through a list of processes running on an operating system of the first computing system; and
determining if any of the processes running on the operating system match a name of the video communication service.
4 . The method of claim 1 , wherein the keyboard sound mitigation model determines if the first computing system is actively running a video communication service according to a timing monitor.
5 . The method of claim 1 , wherein the keyboard sound mitigation operation comprises generating random sounds.
6 . The method of claim 5 , wherein the random sounds are included in an operating system of the first computing system.
7 . The method of claim 1 , wherein the keyboard sound mitigation operation comprises automatically muting a microphone of the first computing system.
8 . The method of claim 1 , wherein determining that the user of the first computing system about to enter the password, or the sensitive information comprises:
scanning the active webpage or window to find password input fields or sensitive information input fields and detecting that a cursor or user input focus has entered the password input field prior to detecting the initial keyboard event, wherein the keyboard sound mitigation operation is not initiated solely based on detecting the password input field or the cursor entering the password input field.
9 . The method of claim 1 , wherein determining that the user of the first computing system is about to enter the password, or the sensitive information comprises:
scanning underlying HTML code of the webpage for statements that are indicative of entering a password.
10 . The method of claim 1 , wherein determining that the user of the first computing system is entering the password, or the sensitive information comprises:
monitoring one or more keyboard events and identifying an initial keystroke in a password input field as a trigger for initiating the keyboard sound mitigation operation, wherein non-character key events are excluded from triggering the keyboard sound mitigation operation.
11 . A non-transitory storage medium having stored therein instructions that are executable by one or more hardware processors to perform operations comprising:
automatically determining, at a keyboard sound mitigation agent that is implemented on a first computing system, if the first computing system is actively running a video communication service during a video communication session with a second computing system;
automatically determining, by a video communication monitor of the keyboard sound mitigation agent, whether a microphone of the first computing system is actively enabled, and enabling a password input monitoring function of the keyboard sound mitigation agent only when both (i) the video communication service is actively running and (ii) the microphone is actively enabled;
in response to determining that the first computing system is actively running the video communication service during the video communication session with a second computing system and that the password input monitoring function is enabled, automatically determining, by the keyboard sound mitigation agent, that a user of the first computing system is entering a password or sensitive information into a webpage or window that is active on the first computing system, wherein determining that the user is entering the password or sensitive information comprises (i) scanning the active webpage or window to detect a password input field or sensitive information input field, (ii) detecting that a cursor or user input focus has entered the password input field or sensitive information input field, and (iii) detecting entry of an initial keyboard event in the password input field or sensitive information input field; and
in response to determining that the user of the first computing system is entering the password or the sensitive information, performing, by the keyboard sound mitigation agent, a keyboard sound mitigation operation that masks or removes any keyboard sounds made while entering the password or the sensitive information, wherein the keyboard sound mitigation operation is initiated upon detection of the initial keyboard event and is terminated upon detecting that the cursor or user input focus exits the password input field or sensitive information input field, and wherein the keyboard sound mitigation operation is performed only while the microphone remains actively enabled.
12 . The non-transitory storage medium of claim 11 , further comprising:
in response to determining that the first computing system is actively running the video communication service during the video communication session with the second computing system, and before determining that the user of the first computing system is entering the password or the sensitive information, automatically determining, by the keyboard sound mitigation agent, if a microphone of the first computing system is actively turned on, and enabling or disabling a password input monitoring function of the keyboard sound mitigation agent based on whether the microphone is actively turned on, wherein disabling the password input monitoring function prevents scanning of the active webpage or window and monitoring of keyboard events.
13 . The non-transitory storage medium of claim 11 , wherein determining if the first computing system is actively running the video communication service during the video communication session with the second computing system comprises:
iterating through a list of processes running on an operating system of the first computing system; and
determining if any of the processes running on the operating system match a name of the video communication service.
14 . The non-transitory storage medium of claim 11 , wherein the keyboard sound mitigation model determines if the first computing system is actively running a video communication service according to a timing monitor.
15 . The non-transitory storage medium of claim 11 , wherein the keyboard sound mitigation operation comprises generating random sounds.
16 . The non-transitory storage medium of claim 15 , wherein the random sounds are included in an operating system of the first computing system.
17 . The non-transitory storage medium of claim 11 , wherein the keyboard sound mitigation operation comprises automatically muting a microphone of the first computing system.
18 . The non-transitory storage medium of claim 11 , wherein determining that the user of the first computing system about to enter the password, or the sensitive information comprises:
scanning the active webpage or window to find password input fields or sensitive information input fields and detecting that a cursor or user input focus has entered the password input field prior to detecting the initial keyboard event, wherein the keyboard sound mitigation operation is not initiated solely based on detecting the password input field or the cursor entering the password input field.
19 . The non-transitory storage medium of claim 11 , wherein determining that the user of the first computing system is about to enter the password, or the sensitive information comprises:
scanning underlying HTML code of the webpage for statements that are indicative of entering a password.
20 . The non-transitory storage medium of claim 11 , wherein determining that the user of the first computing system is entering the password, or the sensitive information comprises:
monitoring one or more keyboard events and identifying an initial keystroke in a password input field as a trigger for initiating the keyboard sound mitigation operation, wherein non-character key events are excluded from triggering the keyboard sound mitigation operation.