Abstract: The disclosed techniques provide systems and methods for detecting malicious or otherwise abusive access of private end-user data in social networking systems. More specifically, various malicious action detection procedures are described for identifying a target user account that is examined via a private data access, generating a relationship scorecard including various social factors that together indicate a measure of social connectedness between a source user that initiates the private data access and the target user whose account is examined via the private data access, and making a determination as to whether the private data access is potentially abusive based on the social connectedness.