Retrospective learning of communication patterns by machine learning models for discovering abnormal behavior
Conventional email filtering services are not suitable for recognizing sophisticated malicious emails, and therefore may allow sophisticated malicious emails to reach inboxes by mistake. Introduced here are threat detection platforms designed to take an integrative approach to detecting security threats. For example, after receiving input indicative of an approval from an individual to access past email received by employees of an enterprise, a threat detection platform can download past emails to build a machine learning (ML) model that understands the norms of communication with internal contacts (e.g., other employees) and/or external contacts (e.g., vendors). By applying the ML model to incoming email, the threat detection platform can identify security threats in real time in a targeted manner.
1. A method comprising:
establishing, via an application programming interface, a connection with a storage medium that includes information related to communication activities of an enterprise;
downloading, via the application programming interface, a series of communications received by an employee over an interval of time;
providing the series of communications to a machine learning (ML) model as training data, so as to produce a trained ML model that is able to identify deviations in features, content, or context of communications received by the employee;
storing the trained ML model in a profile that is associated with the employee or the enterprise;
generating a statistical profile that includes at least one score by providing at least two attributes of a first communication to the trained ML model as input, wherein each score corresponds to a pair of attributes selected from amongst the at least two attributes, and wherein each score is based on an analysis of the corresponding pair of attributes by the trained ML model; and
determining, based on the statistical profile, whether the first communication represents a security risk, including by comparing each score in the statistical profile to a corresponding threshold that is calibrated based on a threshold at which at least one of false positives or false negatives are to be generated by the trained ML model.
2. The method of claim 1 , further comprising: in response to a determination that the first communication represents the security risk, characterizing the security risk along multiple dimensions.
3. The method of claim 2 , wherein the multiple dimensions include:
an attacked party,
an attack vector,
an impersonated party,
an impersonation strategy, and
an attack goal.
4. The method of claim 1 , wherein the storage medium is managed by an entity that supports an office suite or an email suite that is utilized by employees of the enterprise.
5. The method of claim 1 , further comprising:
downloading, via the application programming interface, a second series of communications received by the employee over a second interval of time that precedes the interval of time; and
applying the trained ML model to the second series of communications to establish whether any communications received during the second interval of time represent a security risk.
6. The method of claim 1 , further comprising: in response to a determination that the first communication poses a security risk, performing a remediation action to mitigate a threat posed by the first communication.
7. The method of claim 1 , wherein the
at least two attributes include (i) a primary attribute extracted from the first communication and (ii) a secondary attribute determined based on the primary attribute and additional information associated with the employee.
8. The method of claim 1 , further comprising updating the statistical profile to reflect changes in a communication behavior of the enterprise.
9. A non-transitory medium with instructions stored thereon that, when executed by a processor of a computing device, cause the computing device to perform operations comprising:
establishing, via an application programming interface, a connection with a storage medium that includes information related to communication activities of an enterprise;
downloading, via the application programming interface, a series of communications received by an employee over an interval of time;
providing the series of communications to a machine learning (ML) model as training data, so as to produce a trained ML model that is able to identify deviations in features, content, or context of communications received by the employee;
storing the trained ML model in a profile that is associated with the employee or the enterprise;
generating a statistical profile that includes at least one score by providing at least two attributes of a first communication to a trained ML model as input, wherein each score corresponds to a pair of attributes selected from amongst the at least two attributes, and wherein each score is based on an analysis of the corresponding pair of attributes by the trained ML model; and
determining, based on the statistical profile, whether the first communication represents a security risk, including by comparing each score in the statistical profile to a corresponding threshold that is calibrated based on a threshold at which at least one of false positives or false negatives are to be generated by the trained ML model.
10. The non-transitory medium of claim 9 , wherein the operations further comprise: in response to a determination that the first communication represents the security risk, characterizing the security risk along multiple dimensions.
11. The non-transitory medium of claim 10 , wherein the multiple dimensions include:
an attacked party,
an attack vector,
an impersonated party,
an impersonation strategy, and
an attack goal.
12. The non-transitory medium of claim 9 , wherein the storage medium is managed by an entity that supports an office suite or an email suite that is utilized by employees of the enterprise.
13. The non-transitory medium of claim 9 , wherein the operations further comprise:
downloading, via the application programming interface, a second series of communications received by the employee over a second interval of time that precedes the interval of time; and
applying the trained ML model to the second series of communications to establish whether any communications received during the second interval of time represent a security risk.
14. The non-transitory medium of claim 9 , wherein the operations further comprise: in response to a determination that the first communication poses a security risk, performing a remediation action to mitigate a threat posed by the first communication.
15. The non-transitory medium of claim 9 , wherein the at least two attributes include (i) a primary attribute extracted from the first communication and (ii) a secondary attribute determined based on the primary attribute and additional information associated with the employee.
16. The non-transitory medium of claim 9 , wherein the operations further comprise: updating the statistical profile to reflect changes in a communication behavior of the enterprise.
17. A system comprising:
a processor, and a memory, configured to:
establish, via an application programming interface, a connection with a storage medium that includes information related to communication activities of an enterprise;
download, via the application programming interface, a series of communications received by an employee over an interval of time;
provide the series of communications to a machine learning (ML) model as training data, so as to produce a trained ML model that is able to identify deviations in features, content, or context of communications received by the employee;
store the trained ML model in a profile that is associated with the employee or the enterprise;
generate a statistical profile that includes at least one score by providing at least two attributes of a first communication to the trained ML model as input, wherein each score corresponds to a pair of attributes selected from amongst the at least two attributes, and wherein each score is based on an analysis of the corresponding pair of attributes by the trained ML model; and
determine, based on the statistical profile, whether the first communication represents a security risk, including by comparing each score in the statistical profile to a corresponding threshold that is calibrated based on a threshold at which at least one of false positives or false negatives are to be generated by the trained ML model; and
a memory coupled to the processor and configured to provide the processor with instructions.
18. The system of claim 17 , wherein the processor is further configured to: in response to a determination that the first communication represents the security risk, characterize the security risk along multiple dimensions.
19. The system of claim 18 , wherein the multiple dimensions include:
an attacked party,
an attack vector,
an impersonated party,
an impersonation strategy, and
an attack goal.
20. The system of claim 17 , wherein the storage medium is managed by an entity that supports an office suite or an email suite that is utilized by employees of the enterprise.
21. The system of claim 17 , wherein the processor is further configured to:
download, via the application programming interface, a second series of communications received by the employee over a second interval of time that precedes the interval of time; and
apply the trained ML model to the second series of communications to establish whether any communications received during the second interval of time represent a security risk.
22. The system of claim 17 , wherein the processor is further configured to: in response to a determination that the first communication poses a security risk, perform a remediation action to mitigate a threat posed by the first communication.
23. The system of claim 17 , wherein the at least two attributes include (i) a primary attribute extracted from the first communication and (ii) a secondary attribute determined based on the primary attribute and additional information associated with the employee.
24. The system of claim 17 , wherein the processor is further configured to update the statistical profile to reflect changes in a communication behavior of the enterprise.