The present disclosure relates to techniques for helping targeted users determine whether it is safe to supply personal information requested by a web site. In one embodiment, a method generally includes extracting textual content from a web page that requests information from a user and determining, based on the textual content, the type of information requested. A service type the web page provides is also determined based on the textual content. The service type and the information type are then compared to a set of predefined rules to determine a risk level associated with the web page. A visual indicator of the risk level is then displayed with the web page.