Google’s AI video classifiers are easily fooled by subliminal images by James Vincent @jjvincent
" ... A recent research paper, published by the University of Washington and spotted by Quartz, makes the problem clear. It tests Google’s Cloud Video Intelligence API, which is designed to be used by clients to automatically classify the content of videos with object recognition. (The system is in private beta and not currently in use on YouTube or any other Google products.) The API, which is powered by deep neural networks, works very well against regular videos, but researchers found it was easily tricked by a determined adversary. ... "