System and method to safeguarding sensitive information in cobrowsing session
Co-browsing sessions allow an agent to see inputs of a user and assist the user in completing a task associated with the inputs, such as completing a form on an application or website. Agents may see the information provided by the user, unless that information is sensitive (e.g., social security number, account number, password, etc.), in which case the information is blocked. However, humans will make mistakes for any number of reasons. When that mistake is the providing of sensitive information in a non-sensitive field, systems and methods are provided to block such information, even when provided into a field not designated for sensitive information (e.g., city of residence). As a result, sensitive information may be entered during a co-browsing session, into a field by mistake, and not expose the information to the agent.
1. A system, comprising:
a communication interface to a network; and
a processor;
wherein the communication interface receives, via the network, a data entry by a user utilizing a user device; and
wherein the processor selectively presents or blocks the data entry from being presented to an agent by an agent device, while the agent device and the user device are engaged in a co-browsing session by:
applying a temporary mask of the data entry by the user;
determining whether the data entry by the user matches a sensitive data mask;
wherein the processor maintains presentation of the temporary mask of the data entry by the agent device upon the processor determining that the data entry comprises a sensitive data match entered into a non-sensitive data entry field; and
wherein the processor replaces the temporary mask of the data entry by the user and presents the data entry upon the processor determining that the data entry comprises non-sensitive data entered into the non-sensitive data entry field.
2. The system of claim 1 , wherein maintaining the temporary mask blocks the data entry from being presented to the agent comprises presenting, via the agent device, and presents indicia of the data entry without presenting contents of the data entry.
3. The system of claim 2 , wherein the indicia of the data entry comprises a representation of individual characters of the data entry.
4. The system of claim 1 , wherein the processor determines the data entry comprises sensitive data upon matching the data entry to a pattern known to indicate sensitive data type.
5. The system of claim 4 , wherein the processor provides a prompt to at least one of the user or the agent indicating the sensitive data type.
6. The system of claim 1 , wherein the processor determines whether the data entry comprises sensitive data upon, further comprising executing a neural network trained to recognize sensitive data.
7. The system of claim 6 , wherein training the neural network comprises:
accessing a set of data entries that comprise sensitive data;
applying one or more transformations to the set of data entries including, character alteration to a different character of the same character type, white space insertion, white space removal, punctuation insertion, and punctuation removal to create a modified set of data entries;
creating a first training set comprising the accessed set of data entries, the modified set of data entries, and a set of non-sensitive data entries;
training the neural network in a first stage using the first training set;
crating a second training set for a second stage of training comprising the first training set and the set of non-sensitive data entries that are incorrectly detected as data entries that comprise sensitive data in the first stage of training; and
training the neural network in a second stage using the second training set.
8. The system of claim 1 , wherein the agent device and the user device are engaged in a voice communication via a voice channel concurrently with the co-browsing session.
9. The system of claim 1 , wherein the agent device and the user device are engaged in a text communication via a text channel concurrently with the co-browsing session.
10. A method, comprising:
conducting a co-browsing session over a network with a user device utilized by a user and an agent device utilized by an agent;
receiving an input from a user into a data entry field; and
selectively maintaining or removing a temporary mask of the input to the agent via the agent device;
wherein presentation of the temporary mask is maintained to block the input on the agent device upon determining that the data entry comprises a sensitive data match entered into the data field having a non-sensitive data field identifier; and
wherein the temporary mask is replaced by the input determining that the data entry comprises non-sensitive data entered into the data field having the non-sensitive data field identifier.
11. The method of claim 10 , wherein maintaining the temporary mask blocks the data entry from being presented to the agent comprises presenting, via the agent device, and presents indicia of the data entry without presenting the contents of the data entry.
12. The method of claim 11 , wherein the indicia of the data entry comprises a representation of individual characters of the data entry.
13. The method of claim 10 , further comprising determining the data entry comprises sensitive data comprising matching the data entry to a pattern known to indicate sensitive data type.
14. The method of claim 13 , further comprises prompting at least one of the user or the agent the prompt indicating the sensitive data type.
15. The method of claim 10 , further comprising determining that the data entry comprises sensitive data upon executing a neural network trained to recognize sensitive data.
16. The method of claim 15 , wherein the neural network is trained, the training comprising:
accessing a set of data entries that comprise sensitive data;
applying one or more transformations to the set of data entries, including, character alteration to a different character of the same character type, white space insertion, white space removal, punctuation insertion, and punctuation removal to create a modified set of data entries;
creating a first training set comprising a collected set of data entries, the modified set of data entries, and a set of non-sensitive data entries;
training the neural network in a first stage using the first training set;
creating a second training set for a second stage of training comprising the first training set and the set of non-sensitive data entries that are incorrectly detected as data entries that comprise sensitive data in the first stage of training; and
training the neural network in a second stage using the second training set.
17. The method of claim 10 , further comprising conducting a voice communication, via a voice communication channel, between the agent and the user concurrently with the co-browsing session.
18. The method of claim 10 , further comprising conducting a text communication, via a text communication channel, between the agent and the user concurrently with the co-browsing session.
19. A server, comprising:
a network interface to a network and interconnecting a user device and an agent device for a co-browsing session, wherein a user provides data to an application executing on the user device;
a processor, comprising a number of microprocessors;
a data storage comprising a non-transitory data storage accessible to the processor; and
wherein the processor performs:
receiving a data entry into the application;
applying a temporary mask to the data entry by the user;
determining whether the data entry by the user matches a sensitive data mask;
replacing the temporary mask of the data entry by the user and presenting the data entry to the agent device upon determining that the data entry omits sensitive data and the data entry is to a non-sensitive data field of the application; and
maintaining presentation of the temporary mask of the data to block the data entry from being presented by the agent device upon determining that the data entry is to the non-sensitive data field and comprises sensitive data.
20. The server of claim 19 , wherein the processor provides the temporary mask of the data comprising presenting indicia of the data entry that omits contents of the data entry when presenting the temporary mask of the data to block the data entry from the agent device.