MIT’s Computer Science and Artificial Intelligence Lab (CSAIL) permanently pulls its highly cited 80 Million Tiny Images dataset, after researchers found offensive content and labelling. Scraped from Google Images in 2008, the photo library was created for training computer vision systems in advanced object-detection techniques. “It is clear that we should have manually screened [the images],” CSAIL’s Antonio Torralba told The Register. “For this, we sincerely apologize.”