SynthID Detector Exposes Hidden Watermark in AI Content

Crypto News AI

Google’s SynthID Detector Exposes Hidden Watermark in AI-Generated Content

Written by Aaron S.,

Editor-In-Chief

Last Updated: May 21, 2025

Key Takeaways

Google’s SynthID Detector reveals AI-generated content by spotting hidden watermarks in text, images, video, and audio;
The tool launched as schools face rising cases of students using AI tools like ChatGPT for assignments;
SynthID Detector marks AI-generated text by subtly adjusting word choices without changing meaning or readability.

Stop overpaying - start transferring money with Ogvio. Sign up, invite friends & grab Rewards now! 🎁

Google’s SynthID Detector Exposes Hidden Watermark in AI-Generated Content

BITDEGREE IN YOUR SOCIAL FEED

On May 20, Google’s DeepMind introduced a new tool to help people identify whether content, such as text, images, audio, or video, was made using its artificial intelligence (AI) systems.

The tool, called SynthID Detector, looks for hidden watermarks built into content created by Google’s AI models. These watermarks can be picked up by the tool to show that something was made with AI.

This launch comes at a time when educators are dealing with more students using AI to complete assignments. According to a report from New York Magazine, a professor at Santa Clara University asked students to write personal reflections, only to discover that one had used ChatGPT instead.

5 Best Crypto Portfolio Diversification Strategies (Animated Explanation)

Did you know?

Want to get smarter & wealthier with crypto?

Subscribe - We publish new crypto explainer videos every week!

Google first introduced SynthID in August 2023 to detect AI-generated images. SynthID Detector expands that idea to audio, video, and written content.

It works across Google’s AI tools, including Imagen, Lyria, and text tools inside NotebookLM and Gemini AI.

In the announcement, Google explained:

For text, SynthID looks at which words are going to be generated next, and changes the probability for suitable word choices that wouldn't affect the overall text quality and utility.

They added that if a passage includes many of these chosen word patterns, it will likely be detected as AI-written.

Meanwhile, a journalist at Android Authority recently discovered that Google's Gemini AI works during flights even without paid Wi-Fi. How? Read the full story.