A software developer claims to have reverse-engineered Google DeepMind\u2019s SynthID system, showing how AI watermarks can be stripped from generated images or manually inserted into other works. A claim that, according to Google, isn\u2019t true.<\/p>\n
The developer, going by the username Aloshdenny<\/a>, has open-sourced their work on GitHub<\/a> and documented his process, claiming all it required was 200 Gemini-generated images, signal processing, and \u201cway too much free time.\u201d A little weed also seemed to help.<\/p>\n \u201cNo neural networks. No proprietary access,\u201d Aloshdenny said on Medium<\/a>. \u201cTurns out if you\u2019re unemployed and average enough \u2018pure black\u2019 AI-generated images, every nonzero pixel is literally just the watermark staring back at you.\u201d<\/p>\n SynthID is a near-invisible watermarking system that tags content generated by Google\u2019s AI tools, embedding itself in the pixels of images at the point of creation. It was designed to be difficult to remove without degrading the image quality, and is used widely across the AI products offered by Google \u2014 everything spat out by models like Nano Banana and Veo 3 carries SynthID watermarks, and it\u2019s even being applied to YouTube\u2019s AI-generated creator clones<\/a>.<\/p>\n Aloshdenny says he found the system to be \u201cgenuinely good engineering,\u201d and was still unable to remove SynthID entirely in tests, instead relying on confusing SynthID decoders that try to read watermarked images.<\/p>\n The process used to crack the underlying mechanics of Google\u2019s watermark is technically complex for non-developers. You can read the full breakdown on Aloshdenny\u2019s Medium page (which was apparently written up while Aloshdenny was \u201chigh\u201d) if you\u2019re curious, but here\u2019s a simplified explainer:<\/p>\n \u201cThe fact that the best I could pull off was confuse the decoder enough that it gives up \u2014 not actually delete the thing \u2014 says a lot about how well it was designed,\u201d says Aloshdenny. \u201cIt\u2019s not perfect. But it\u2019s not trying to be unbreakable. It\u2019s trying to raise the cost of misuse high enough that most people don\u2019t bother.\u201d<\/p>\n I haven\u2019t tried Aloshdenny\u2019s project that reverse-engineers Google\u2019s SynthID watermarking system, so I can\u2019t vouch for how effective it actually is. That said, at this point in time, it doesn\u2019t appear that SynthID has been reverse-engineered, at least not to the point where script-kiddies can download a tool and remove (or add) Google\u2019s watermark to trick AI detection systems. Google also doesn\u2019t believe it stands up to Aloshdenny\u2019s claims.<\/p>\n \u201cIt is incorrect to say this tool can systematically remove SynthID watermarks,\u201d Google spokesperson Myriam Khan told The Verge. \u201cSynthID is a robust, effective watermarking tool for AI-generated content.\u201d<\/p>\n","protected":false},"excerpt":{"rendered":"A software developer claims to have reverse-engineered Google DeepMind\u2019s SynthID system, showing how AI watermarks can be stripped…\n","protected":false},"author":2,"featured_media":5091,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[24,25,132,66,781],"class_list":{"0":"post-5090","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-ai","8":"tag-ai","9":"tag-artificial-intelligence","10":"tag-google","11":"tag-news","12":"tag-tech"},"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/5090","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/comments?post=5090"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/5090\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media\/5091"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media?parent=5090"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/categories?post=5090"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/tags?post=5090"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}