In some examples, the format that obscures underlying content may include an image format. The image format may enable an image to leak sensitive data by displaying text without the text being computer-readable as a string. In the example of FIG. 4, application 230 may attempt to output data object 212, which may include an image 402 that displays sensitive data (e.g., a Social Security number) in an obfuscated manner. In this example, image 402 may display the number in a warped or CAPTCHA-style format that constitutes a barrier to optical character recognition of the image. In other examples, image 402 may display the sensitive data in a straight and uniform, or non-warped, manner that nevertheless requires optical character recognition or other analysis to extract the sensitive information as distinct from directly reading the string of text in the same manner that applications read strings of text from document files. In other words, image 402 may encode the sensitive data indirectly by encoding the sensitive data as pixels, sound waves, or other multimedia attributes as distinct from encoding the sensitive data directly in terms of character codes or text strings (e.g., ASCII characters). In other examples, an encrypted format may also enable the data object or corresponding document to leak sensitive data by preventing a data loss prevention system from detecting the sensitive data due to encryption.
