KI oder nicht KI?: SynthID: Google öffnet KI-Detektor für alle Nutzer
Das kann nicht echt sein, oder? Immer bessere KI-Modelle machen es schwierig, künstlich erstellte Inhalte aufzuspüren. Mit Googles SynthID-Detektor soll das leichter werden - mit Einschränkungen.
Google has made its KI-Detektor, known as SynthID, available to all users. This tool allows users to check images, videos, or audio files for possible creation with Google's AI. SynthID looks for invisible digital watermarks embedded in AI content. The detector can identify content from Google's own AI models Gemini, Imagen, Lyria, and Veo, as well as content from partners like OpenAI (ChatGPT), Nvidia, Kakao, and soon Apple.
It's important to note that SynthID only identifies the presence of a watermark, not whether the content is definitively AI-generated. Not all AI-generated content is marked with digital watermarks like SynthID. Since its introduction in 2023, Google has tagged over 180 billion images, videos, or audio files with these watermarks.
OpenAI, the operator of ChatGPT, has also developed a similar tool, which can be used to examine files for hints that they were created using OpenAI's tools. However, a negative result does not necessarily mean that a file was not created with the help of AI.
Rewritten by urgent.news; may contain errors — check the original.