Google will now tell users if an image has been generated using AI, here's how

SynthID

Google’s DeepMind has finally made its AI watermarking technology SynthID open-source. The tool was initially designed to identify AI-generated files and is now accessible via Google Responsible Generative AI Toolkit. By making SynthID freely available, Google aims to democratize access to advanced watermarking tools to identify AI-generated content without altering its visible features.

It is believed to be a significant step toward enhancing the safety, transparency, and traceability of AI-generated content so that people have greater trust in all the content visible on the internet. Google has also successfully integrated SynthID into its Gemini chatbot without compromising the quality, accuracy, creativity, or speed of the generated text.

The best part about SynthID is that it works without human intervention and works by subtly modifying the probability of text outputs in a way that this software can detect but humans cannot. It further adds an imperceptible watermark to AI-generated text using advanced deep-learning models.

To detect the watermark and determine whether the text or the media file has been generated by an AI tool, SynthID compares the expected probability scores for words in the watermarked and unwatermarked files and with a unique pattern of these adjusted scores creates the watermark.

Google’s decision to open-source SynthID for AI text watermarking represents a significant step towards a new era of transparency in the evolving digital landscape.