Watermarking AI-generated content could become the new norm


Google DeepMind has recently shared that it's open-sourcing its SynthID text watermarking tool.

SynthID is a tool for watermarking content generated using artificial intelligence (AI), including video, images, and text.

Google DeepMind, an AI research laboratory, has announced that it's making its tool accessible to developers and businesses to help them identify AI-generated content.

ADVERTISEMENT

The digital watermark, which can't be seen with the naked eye, can withstand various editing techniques, such as trimming, reordering, adding noise, cropping, and filters.

In images, the watermark is embedded in the pixels, while in video, it's used in every frame. For audio, a signal is converted into a spectrogram, or representation of soundwaves, which then gets watermarked.

The tool can be used not only for watermarking but also for detecting AI-generated content.

Previously, Google announced that SynthID is being applied to the Gemini AI chatbot, showing that watermarking helps detect content generated by AI, some of which can pose risks by being misinformative or unsafe.

"​​By open-sourcing the code, more people will be able to use the tool to watermark and determine whether text outputs have come from their own LLMs - making it easier to build AI responsibly," shared Google DeepMind.

ADVERTISEMENT