English 
Google launched a verification portal called SynthID Detector to help users identify content generated using artificial intelligence (AI). The site allows users to upload content and identifies whether the file contains the SynthID watermark that is automatically embedded in media generated by Google’s AI tools. The site can identify areas in photos that are likely to have been altered by AI. Google recently added SynthID detection to Google Photos, which shows when photos have been altered with the company’s Magic Editor. The new portal will perform similar detection on a wider range of media, including text and images generated by Gemini, videos generated by Veo, and audio generated by Lyria. Media generated using Google’s AI tools will automatically be embedded with an invisible SynthID watermark. The watermark is designed to be resistant to basic digital manipulation and remains detectable even when shared on social media or messaging apps.
検出可能なコンテンツの制限このポータルは、ディープフェイクのようなAIが生成した誤情報からユーザーを保護し、オリジナル作品と合成メディアを区別するための重要な第一歩です。しかし、ChatGPTのようなSynthID透かしを埋め込んでいないツールによって生成されたコンテンツの検出には対応していません。Googleは、今年3月にNVIDIAとの提携を発表するなど、自社のツール以外にもSynthIDの利用を拡大するために外部と協力しています。また、コンテンツ検証会社GetReal社との新たな提携も発表しており、同社の検証ツールにSynthID検出機能を追加する予定です。
Limits of digital watermarkingHowever, it is impossible to fight deepfakes and AI-generated misinformation using digital watermarks alone. Open source AI tools will continue to exist, and it will be difficult to force them to embed watermarks. Although it may still feel easy to spot AI-generated content at present, that is not necessarily the case in the future. As generative AI becomes more sophisticated, tools like SynthID are likely to be crucial in distinguishing between human-created content and what AI has produced. Google’s SynthID Detector portal is already open to a select group of testers, with a waiting list for media and researchers.On May 20, Google launched a verification portal called SynthID Detector to help users identify content generated using artificial intelligence (AI). The site allows users to upload content and identify whether the file contains the SynthID watermark that is automatically embedded in media generated by Google’s AI tools. The site can identify areas in photos that are likely to have been altered by AI. Google recently added SynthID detection to Google Photos, which shows when photos have been altered with the company’s Magic Editor. The new portal will perform the same detection on a wider range of media, including text and images generated by Gemini , videos generated by Veo, and audio generated by Lyria. Media generated using Google’s AI tools will automatically be embedded with an invisible SynthID watermark. The watermark is designed to be resistant to basic digital manipulation and remains detectable even when shared on social media or messaging apps.
こちらもお読みください： Invox社、CO2排出量追跡のためのAIツールを発表
検出可能なコンテンツの制限このポータルは、ディープフェイクのようなAIが生成した誤情報からユーザーを保護し、オリジナル作品と合成メディアを区別するための重要な第一歩です。しかし、ChatGPTのようなSynthID透かしを埋め込んでいないツールによって生成されたコンテンツの検出には対応していません。 グーグル は、今年3月にNVIDIAとの提携を発表するなど、自社のツール以外にもSynthIDの利用を拡大するために外部と協力しています。また、コンテンツ検証会社のGetReal社との新たな提携も発表しており、同社の検証ツールにSynthIDの検出機能を追加する予定です。
Limits of digital watermarkingHowever, it is impossible to fight deepfakes and AI-generated misinformation using digital watermarks alone. Open source AI tools will continue to exist, and it will be difficult to force them to embed watermarks. Although it may still feel easy to spot AI-generated content at present, that is not necessarily the case in the future. As generative AI becomes more sophisticated, tools like SynthID are likely to be crucial in distinguishing between human-created content and what AI has produced. Google’s SynthID Detector portal is already open to a select group of testers, with a waiting list for media and researchers.
ソース ヤフー
日本語