For this vulnerability to be exploitable, the ImageMagick image processing library needs to be used to resize or add a text watermark to a user-uploaded file which was saved using a user-provided filename, or where the parameters for adding a watermark are user-controlled. File upload implementations that use a randomly generated filename before image resizing are not vulnerable.
This vulnerability is simple to exploit and we expect to see active exploitation soon. However, attackers will need to locate file upload functionality within your applications first which will be difficult to fully automate at scale, so mass exploitation is unlikely.