[analyse_image type=”featured” src=”https:\/\/www.artnews.com\/wp-content\/uploads\/2026\/08\/MIT-CSAIL-study-AI-generated-images.jpeg?w=1024″]<\/p>\n
\n\tA new study from researchers at MIT\u2019s Computer Science and Artificial Intelligence Laboratory (CSAIL) suggests that the more data a generative model is trained on, the less likely an AI-generated image is to be traced back to a source image. They call this phenomenon attribution decay.<\/p>\n
\n\t\u201cIf you take away a piece of data and the output of the model doesn\u2019t change, then that piece of data didn\u2019t affect the output,\u201d said Zheng Dai, a former MIT CSAIL researcher and the lead author on the study, published yesterday in\u00a0Nature Communications<\/em>. \u201cSo it doesn\u2019t make much sense to attribute the output to that piece of data. And if you then do this one at a time for every other piece of data and find that the output doesn\u2019t change for any of them either, then it doesn\u2019t make much sense to attribute the output to any one of them.\u201d<\/p>\n \n\tThe researchers trained 24 ensembles on datasets ranging from 256 images to more than 160,000, taken from public collections. They say their method involved retraining the model from scratch each time they removed a particular image so that they could observe concrete results; previous studies, they say, relied on approximations to estimate the influence of any one source image. They built an architecture they call a \u201cdiffusion ensemble,\u201d with several models, each trained on a different data subset.<\/p>\n \n\t\u201cAll previous methods were approximate,\u201d said MIT professor David Gifford, a CSAIL principal investigator. \u201cThey really could not absolutely show that deleting individual things did not change the output. This paper introduces the first method that is absolute. You\u2019re actually deleting the inputs and deleting all influences of the inputs. This is the first exact method for doing large-scale deletion efficiently and showing that the results don\u2019t change.\u201d<\/p>\n \n\tThe authors put their ensemble up against 24 conventional diffusion models trained on the same data and found that the images \u201ccame out looking about as good by standard measures,\u201d according to press materials.\u00a0<\/p>\n \n\tGifford believes the results bear on the legal question of whether models\u2019 outputs are merely derivative works, with implications for copyright.<\/p>\n \n\t\u201cOne way to think about this is that these models are creative,\u201d he said in press materials. \u201cThey are not simply copying what they are fed, but creating brand new outputs. If those outputs have nothing to do with any individual piece of training data, that raises questions about fair use, about whether the outputs are themselves copyrightable as novel works, and about how authors get compensated when what comes out of a model isn\u2019t attributable to anything on the internet.\u201d<\/p>\n<\/div>\n<\/div>\n [analyse_source url=”https:\/\/www.artnews.com\/art-news\/news\/mit-study-ai-generated-images-trace-back-sources-1234795260\/”]<\/p>\n","protected":false},"excerpt":{"rendered":" [analyse_image type=”featured” src=”https:\/\/www.artnews.com\/wp-content\/uploads\/2026\/08\/MIT-CSAIL-study-AI-generated-images.jpeg?w=1024″] A new study from researchers at MIT\u2019s Computer Science and Artificial Intelligence Laboratory (CSAIL) suggests that the more data a generative model is trained on, the less likely an AI-generated image is to be traced back to a source image. They call this phenomenon attribution decay. \u201cIf you take away a piece […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[61,226],"class_list":["post-2104201","post","type-post","status-publish","format-standard","hentry","category-politics","tag-artnews-com","tag-crawlmanager"],"_links":{"self":[{"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=\/wp\/v2\/posts\/2104201","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=2104201"}],"version-history":[{"count":0,"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=\/wp\/v2\/posts\/2104201\/revisions"}],"wp:attachment":[{"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=2104201"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=2104201"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=2104201"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}