I'm a software developer who has been working on a yet-to-be released content filtering application. One of the filtering categories is of course pornography, and once upon a time I decided I was going to develop a pornographic image detector/classifier. That is to say, the program would load the image, perform some analysis of the image and classify it accordingly.
Anyone who has ever developed anything like this knows that you need a significant base of source material from which to derive your original classification models. So, I installed DownloadThemAll into Firefox and started punching in some nasty terms plus the words "Tumblr". Material collected.
I ended up having thousands of images, both pornographic and non-pornographic so that the algorithm could extract features to make accurate classifications. However, after some time of developing and having never actually personally analyzed these thousands of images, my wife and I discussed the very real potential that something unspeakable may have been on one of these sites that I scripted to have all images dragged down from. So, I ceased development destroyed everything I had collected, deciding the risk of inadvertently having such filth was definitely not worth it.
So my question is, would I be liable if something had of been there? How can I go about such research while protecting myself (if at all)? The last thing on earth I'd want in life is being thrown in a headline next to Jared Fogle, with the only excuse being "I didn't know! It was for science!"
Google actually copies and stores image results on their own servers. So they can and do have instances where illegal explicit material is copied and stored on private servers. How are they covered from liability or possession?