{"id":4210,"date":"2026-07-22T12:14:31","date_gmt":"2026-07-22T12:14:31","guid":{"rendered":"https:\/\/silvybrand.com\/?p=4210"},"modified":"2026-07-22T12:14:31","modified_gmt":"2026-07-22T12:14:31","slug":"meta-ai-detection-labeling-content-seal-watermarks-synthid","status":"publish","type":"post","link":"https:\/\/silvybrand.com\/?p=4210","title":{"rendered":"Meta made its own AI detection system. It should have just used Google\u2019s"},"content":{"rendered":"


\n<\/p>\n

\n
\n

IIn March, Meta\u2019s Oversight Board called on the company<\/a> to \u201cmeet its public commitments and employ its own tools\u201d to help quell the spread of deceptive generative AI content across platforms. Meta responded in July by introducing Content Seal<\/a> \u2014 an invisible watermarking technology that flags images generated by the company\u2019s new AI model. But it was basically a footnote buried in the company\u2019s announcement for its Muse image and video generation tools.<\/p>\n<\/div>\n

\n

As someone who spends a lot of time scrutinizing AI labeling systems, Content Seal doesn\u2019t fill me with confidence. There are already more established solutions, like C2PA Content Credentials and Google\u2019s SynthID<\/a>, that Meta could have used instead of launching its own system significantly later. After digging around to figure out why Meta has launched its own system, I\u2019m not convinced that Meta has thought this through.<\/p>\n<\/div>\n

\n

By Meta\u2019s description, Content Seal works similarly to SynthID<\/a>. The watermark, invisible to human eyes, provides a \u201chidden provenance signal\u201d embedded into AI-generated images that can then be scanned and flagged by a detection tool, helping online users to differentiate deepfakes from authentic content. Like SynthID, Meta also says that Content Seal watermarks remain intact and can still be detected if the image is \u201ccropped, compressed, resized, or screenshotted.\u201d<\/p>\n<\/div>\n

\n

So, if Content Seal functionally does the same thing\u2026 why not just adopt SynthID? Meta already operates as a steering committee member<\/a> of the Coalition for Content Provenance and Authenticity (C2PA) that promotes the separate Content Credentials standard alongside<\/em> Google, so it\u2019s shown willingness to work with others on solving the growing issue of AI detection. SynthID has also already been adopted by OpenAI, so clearly Google is also willing to open its technology up to rival AI providers in the name of improving transparency.<\/p>\n<\/div>\n

\n

Content Seal has several limitations in its current state despite those similarities to Google\u2019s system. For now, users can only detect Content Seal watermarks through a dedicated web tool<\/a> that Meta is testing, meaning Meta hasn\u2019t built those detection capabilities into its Meta AI chatbot like Google has with Gemini<\/a>. It sounds like that may be in the works, however. Meta spokesperson Faith Eischen told The Verge<\/em> that the company is \u201cexploring ways to bring detection closer to where people encounter AI-generated content.\u201d Given that\u2019s where AI detection is needed most, and has been for some time, why isn\u2019t it available at launch?<\/p>\n<\/div>\n

\n

The watermark itself is also only being applied to images generated by Muse in the Meta AI app and Meta.ai website, which means online users can\u2019t use it to detect content created by Meta\u2019s older AI models. Support for generated video isn\u2019t available either, though Meta says this is coming \u201csoon.\u201d<\/p>\n<\/div>\n

\n
\n
\n
\n
\n
\"<em>Here\u2019s<\/a><\/div>\n
\"<em>And<\/a><\/div>\n