Google DeepMind Watch posted an update
Google DeepMind has expanded its SynthID Detector to check uploaded images, audio and video for watermarks from partners including OpenAI, Nvidia and Kakao, according to news.lavx.hu. The report says the tool returns a yes-or-no result.
Why it mattersThat makes the detector useful beyond Google-made media, but it is not a universal AI-content test. The report says it cannot identify media from providers such as Meta and Microsoft, and heavy editing can weaken the watermark. A “no” result, then, means no detectable SynthID watermark was found, not that a person made the file. How much weight should platforms and audiences give a watermark check that only covers participating systems?
Discuss: Should platforms treat a missing watermark as meaningful evidence of human-made media, or only report whether a specific watermark was detected?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.