High-quality deepfakes are now almost impossible to distinguish with the naked eye; studies indicate we only get one out of every four fake videos right. Faced with this reality, Google is betting on SynthID, a technology that doesn't try to hunt down the fake, but rather certifies the origin of content from its creation, offering a scalable and reliable defense in the digital ecosystem. 🤖
Expanded verification: C2PA, Search, and API in Google Cloud 🛡️
Google has announced the integration of C2PA Content Credentials into Gemini, allowing tracking of the editing history of images and videos. Additionally, the SynthID detector expands to Search to mark AI-generated results, and a new API in Google Cloud will allow companies to analyze suspicious content. These tools aim to create a standard of transparency without relying on human perception.
Detecting fakes: now with Google's help, not your eye 👁️
It turns out our superpower for detecting deepfakes only works one out of every four attempts. Google gives us a hand with SynthID, even if it's a bit humiliating to admit that an algorithm beats us at the game of spot the difference. But hey, as long as we don't have to pass an eye exam every time we see a viral video, robotic help is welcome.