System and method for computing an organization's insurance underwriting score for cyber insurance using AI
Evaluating an organization's risk to be exposed to cyber security events, by collecting security-based risk indicators about multiple organizations and an indication on whether or not each of the multiple organizations suffered a security event, inputting values of the security-based risk indicators about the multiple organization into a software model, running the software model multiple times, where the software model outputs multiple sets of one or more classifiers, where each set of one or more classifiers distinguishes organizations that suffered the security event from organizations that did not suffer the security event, where the one or more classifiers include classifiers selected from the security-based risk indicators, assigning a risk score for a specific organization according to the importance of classifiers related to the specific organization, and computing an underwriting score for the specific organization.
1. A computerized method for evaluating an organization's risk to be exposed to cyber security events, the method comprising:
collecting security-based risk indicators about multiple organizations and an indication on whether or not each of the multiple organizations suffered a security event;
inputting values of the security-based risk indicators about the multiple organization into a software model;
running the software model multiple times, the software model outputs multiple sets of one or more classifiers, each set of one or more classifiers distinguishes organizations that suffered the security event from organizations that did not suffer the security event, wherein the one or more classifiers comprise classifiers selected from the security-based risk indicators;
assigning a risk score for a specific organization according to the importance of specific classifiers related to the specific organization; and
computing an underwriting score for the specific organization, wherein said computing comprises
comparing a type-based risk score for the specific organization to a type-based benchmark value representing a standard risk score in a specific risk type of multiple other organizations of a reference group having similar characteristics with the specific organization,
assigning a risk-based value based on the risk-based comparison,
accumulating multiple risk-based values for a group of relevant risk types, and
assigning a weight to classifiers outputted by the software model according to the number of times each classifier appeared in a set of the multiple sets outputted by the software model, wherein the weights represent an impact of a specific organization to be exposed to a security event.
2. The method of claim 1 , wherein the risk-based value is assigned based on a range of optional values, the range corresponds to the importance of the specific classifiers related to the specific organization.
3. The method of claim 1 , wherein the risk-based value increases when the distance between the type-based risk score for the specific organization to a type-based benchmark value increases.
4. The method of claim 3 , wherein the risk-based value increases in multiple predefined offset values, each of the predefined offset values is related to a range of distances between the type-based risk score for the specific organization to a type-based benchmark value increases.
5. The method of claim 1 , wherein computing the underwriting score comprises assigning values that represent critical key findings of the specific organization.
6. The method of claim 1 , further comprises outputting an underwriting recommendation to an insurance company based on the underwriting score and the risk score, said underwriting recommendation is selected from a closed group of recommendations.
7. The method of claim 1 , wherein the software model is a random forest model.
8. The method of claim 1 , further comprising:
running the model for multiple different security event types;
identifying the classifiers that are relevant for each security event type of the multiple different security event types.
9. The method of claim 8 , further comprising assigning a weight for classifiers relevant for each security event types of the multiple different security event types.
10. The method of claim 9 , further comprising computing a relative security event rank for a specific organization, said relative security event rank defining a risk of the specific organization to suffer from a specific security event based on the values of the security-based risk indicators of the specific organization relative to other organizations.
11. The method of claim 10 , further comprising computing a total rank for the specific organization according to the relative security event ranks computed for the specific organization for various event types and a weight of each security event type of the multiple event types.
12. The method of claim 1 , further comprising generating a data record of each organization comprising values for the security-based risk indicators and inputting the multiple records into the model.
13. The method of claim 12 , further comprising adjusting the record of each organization according to known vulnerabilities of the values for the security-based risk indicators.
14. The method of claim 12 , wherein the record further comprises non-security risk indicators of the organization.
15. The method of claim 1 , wherein the security-based risk indicators are unique to each organization.
16. The method of claim 1 , wherein the security-based risk indicators comprise security vulnerabilities of the organization.
17. The method of claim 1 , wherein the security-based risk indicators comprise technologies used by the organization.