Environment:
Our project allows teachers to create learning objects (activities, lessons, videos) that may incorporate uploaded images for the purpose of language learning. Our EULA states that users certify that they have the rights to use and distribute the images they upload. Using some JavaScript the activities become part of any web page any number of students may be reading. These images are not actually published from our site or statically available through a search engine. They are only seen when a student elects to dynamically generate activities.
Proposal:
Rather than just allow teachers to upload images, we would like to build a feature that does an Internet image search and makes those images available for use in activities. i.e. Search for "apple" and get pictures of apples you can drag into an activity. The source of the image would be acknowledged with a link to the source page, and it would be up to the author of the activity which images to choose.
Question:
This feels like copyright infringement, but I'm not so sure. Pinterest, Tumblr and countless others seem to be getting around the liability by offering a mechanism for reporting copyright infringement. What is my actual exposure in the real world? Is the worst I can expect to receive cease and desist notices?
Another part of our system allows users to create their own activities that only they, themselves, would see. Would I need to worry about copyright infringement at all in that case?