Editorial illustration for Google's SynthID Begins Early Trials to Detect AI-Generated Content on Gemini
Google's SynthID Trials Aim to Unmask AI-Generated Content
SynthID on Gemini Tested in Early Trials to Detect AI-Generated Content
Google activated its SynthID watermarking system on the Gemini platform in December. It’s live. I ran its detector against a batch of AI-made images, text, and video—some from Gemini itself, some from other sources.
The tool caught some. It missed others entirely.
Now that SynthID has been rolled out on Gemini, we can put it to test to see how well it performs in figuring out AI-generated content. I'd be testing it on the following tasks to test how well it performs in discerning multimodal AI-generated content: These three will test SynthID's ability to recognize the images it should flag, and how it handles those it shouldn't. Note: I'd be using Gemini App and Google AI Studio for performing different tasks, as Gemini App is currently limited in its features. The Text and Video AI-detection was performed on Google AI Studio.
Here’s what happened. For images bearing its own watermark, SynthID worked well—Google’s docs say that’s by design. Detecting AI text and video was spotty.
The capability is real but uneven. Researchers argue more data will refine it. Today, though, whether it works depends completely on what you feed it.
Related Reading
Browse all Google news →
LLMs & Generative AI ChatGPT's 'Nerdy' tweak rewards goblin metaphors in answers, study finds
LLMs & Generative AI Google tests visual 'magazine-style' UI for Gemini 3 Pro users
LLMs & Generative AI AI Engineers Face Rising Costs, Need New Strategies for Efficiency
Related Topic NVIDIA and Google Cloud let developers scale AI from prototype to production
Market Trends
Related Topic Google's FACTS benchmark shows 70% factuality ceiling across four tests
Research & Benchmarks
From Archives OpenAI says shopping prompts aren't ads, researcher warns they may appear so
From Archives Gemini 3 lets users upload sport videos, using long-context and vision
From Archives Google Maps to get hands-free conversational driving via Gemini
Industry Applications
From Archives OpenAI says GPT-5.2 will debut next week, ahead of Google’s Gemini 3 in internal tests
Common Questions Answered
How does Google's SynthID technology aim to detect AI-generated content on the Gemini platform?
SynthID is designed to help distinguish between human and machine-created content across different media types, including text, images, and video. The technology represents Google's strategic approach to bringing transparency and identification capabilities to synthetic media detection.
What are the key challenges SynthID is trying to address in the current digital landscape?
As synthetic media continues to proliferate across various platforms, identifying AI-generated content has become increasingly complex and challenging. SynthID is Google's attempt to develop a robust tool that can accurately recognize and flag AI-created materials across multiple formats.
What platforms will researchers use to test SynthID's multimodal content detection capabilities?
Researchers will be using both the Gemini App and Google AI Studio to perform different testing tasks for SynthID. These platforms will help evaluate the technology's ability to recognize and flag AI-generated images and content with precision across different media types.
Further Reading
- How we're bringing AI image verification to the Gemini app — Google Official Blog
- SynthID Explained: A Technical Deep Dive into DeepMind's Invisible Watermarking System — Dev.to
- SynthID in 2025: Where Google's Invisible Watermark Shows Up (and Where It Doesn't) — Jesus Iniesta Blog
- What Is Google SynthID Watermarking? Invisible Provenance Signals Explained — Skywork AI Blog
- Did Google's AI fool you? - Digital Digging with Henk van Ess — Digital Digging
