The popularity of Online Social Networks (OSN) have brought up new privacy threats. These threats often arise after users willingly, but unwittingly share their information with a wider group of people than they intended. This paper identifies an important problem with posting on social networks and proposes a unique two phases approach to address the problem. The first phase involves recognition of key phrases in potential posts. These key phrases suggest that the message reveals information that might compromise the safety of the poster or someone else. A set of detection rules have been developed for this purpose and tested with over 16000 real Facebook posts. The detection system success rate was about 85%. The second phase involves automatic placement of friends into sets called Circles of Trust. Potentially messages with sensitive information then can be restricted to just the appropriate Circles of Trust. An approach to do the initial assignment of friends to circles and mechanisms for moving friends between circles as their relationship with the poster changes is also discussed.