In providing visual cues in a document, a server captures search term(s) input by a user and detects a user selection of a document returned based on the search term(s). The server determines a dataset describing the user, where the dataset includes the search term(s), the document, and a set of demographic data of the user. The server maps the dataset to a user group, where the user group is associated with an expertise level for a domain associated with the document. The server assigns the expertise level associated with the user group to the user and maps the expertise level of a risk model for the domain. The risk model is applied to the document to identify the content segments that may pose a risk of harm to a reader with the user's expertise level. The document with the visual cues can then be displayed to the user.