Data collaboration protection

1 . A machine-implemented method of protection data in a data collaboration environment, comprising:

training a machine learning model to classify sensitive content and user profiles of users accessing a computer network, and to determine whether users are authorized to view instances of the sensitive content based on identifying associations between the sensitive content and the user profiles;

receiving an indication that an instance of sensitive content was shared with a respective user;

responsive to the indication that the instance the sensitive content was shared with the respective user:

determining, by the machine learning model, whether the respective user is authorized to view the instance of sensitive content based on the training of the machine learning model and the machine learning model identifying one ore more associations between the instance of sensitive content and the user profile of the respective user;

only when the machine learning model determines that the respective user is authorized to view the sensitive content, allowing the instance of sensitive content to be shared with the respective user, automatically without further user intervention; and

when the machine learning model determines that the respective user was not authorized to view the instance of sensitive content, preventing the instance of sensitive content to be shared with the respective user.

2 . The machine-implemented method of claim 1 , wherein the method further comprises, after allowing the instance of sensitive content to be shared with the respective user:

flagging the respective user for future authorization to the instance of sensitive content;

receiving a confirmation that the respective user should be authorized to view the instance of sensitive content; and

training the machine learning model so that a next time that the sensitive content is shared with the respective user the machine learning model will determine whether the respective user is authorized to view the instance of the sensitive content based in part on the respective user being previously authorized to view the sensitive content.

3 . The machine-implemented method of claim 1 , wherein determining whether the respective user is authorized to view the instance of sensitive content comprises:

determining, by the machine learning model, based on the training of the machine learning model, whether the respective user is authorized to view sensitive content based on a plurality of signals, a first signal comprising identifying whether the respective user previously viewed the instance of sensitive content within a predetermined window of time, and a second signal comprising identifying the one or more associations between the instance of sensitive content and the user profile of the respective user.

4 . The machine-implemented method of claim 3 , wherein a third signal of the plurality of signals comprises:

identifying a semantic association between a name of a group that the respective user is associated with and access to a classification of the sensitive content.

5 . The machine-implemented method of claim 3 , wherein a fourth signal of the plurality of signals comprises:

identifying peers of the respective user that have access to the instance of sensitive content.

6 . The machine-implemented method of claim 3 , further comprising:

classifying the sensitive content using a machine learning model to generate a content classification;

classifying the user profiles using the machine learning model to generate profile classifications;

generating embeddings for the content classification and the profile classification using a language model, wherein the embeddings represent a semantic meaning of the content classification and profile classifications; and

determining a semantic similarity between the content classification and a profile classification of the user profile of the respective user by comparing the embeddings;

wherein the plurality of signals comprises associating the instance of sensitive content with the user profile of the respective user based on the determined semantic similarity.

7 . The machine-implemented method of claim 6 , wherein associating the instance of sensitive content with the user profile of the respective user based on the determined semantic similarity comprises identifying that a department, title or position associated with the respective user is associated with access to certain classifications of the sensitive content.

8 . The machine-implemented method of claim 3 , wherein the plurality of signals includes identifying that the instance of sensitive content was authorized to be viewed by other users in a same group with the respective user within a predetermined window of time.

9 . The machine-implemented method of claim 1 , wherein receiving the indication that the sensitive content was shared with the respective user comprises:

providing the machine learning model with a piece of content currently being shared between the users; and

receiving, from the machine learning model based on the providing, an indication that a first user is sharing the instance of sensitive content with the respective user.

10 . The machine-implemented method of claim 1 , wherein the machine learning model comprises a large language model configured for language processing.

11 . The machine-implemented method of claim 10 , wherein receiving the indication that the sensitive content was shared with the respective user comprises:

identifying a document shared between two or more of the users; and

passing the document to the large language model to identify the sensitive content and to determine whether the respective user is authorized to view the sensitive content.

12 . The machine-implemented method of claim 1 , wherein determining whether the respective user is authorized to view the instance of the sensitive content based on the respective user being previously authorized to view the sensitive content, comprises assigning greater weight to the respective user being previously authorized to view the sensitive content commensurate with how many times the respective user was previously confirmed to be authorized to view the instance of sensitive content.

13 . The machine-implemented method of claim 1 , wherein allowing the instance of sensitive content to be shared with the respective user comprises overriding or bypassing an action performed by a data loss protection system with regard to the sharing of the instance of sensitive content, and wherein preventing the instance of sensitive content to be shared with the respective user comprises allowing the data loss protection system to perform normal operations with regard to the sharing of the instance of sensitive content.

14 . A system, comprising:

a server comprising:

one or more processors; and

a non-transitory memory storing instructions that, when executed by the one or more processors, causes the one or more processors to facilitate performance of the machine-implemented method of claim 1 .

15 . A non-transitory machine readable medium storing instructions thereon that, when executed by a machine, causes the machine to perform the machine-implemented method of claim 1 .