Google has opened its SynthID detection tool to the public, letting anyone upload images, audio, or video files to check whether they were AI-generated. Previously limited to media professionals and journalists, the tool is now freely accessible worldwide.
The timing matters. AI image generators now produce outputs convincing enough to fool most viewers, making it harder than ever to separate authentic content from fabricated material. Google says it has already embedded SynthID watermarks in more than 180 billion images and videos, along with 240,000 years of audio. The rollout brings detection capabilities once reserved for specialists to everyday internet users.
At a Glance
- SynthID Detector is now free and publicly accessible at synthid-detector.google
- Users upload images, audio, or video directly through their browser
- Detects watermarks from OpenAI, NVIDIA, Kakao, and soon Apple
- Cannot detect Meta's Content Seal or unwatermarked AI content
How SynthID Works and Its Limits
SynthID embeds invisible watermarks into AI-generated content at the point of creation. These markers survive common modifications like compression and color adjustments. When you upload a file, SynthID scans for these invisible markers and reports whether AI generation is detected.
The main limitation: content created with AI systems that don't use SynthID watermarking won't register as AI-generated. Meta's Content Seal, for instance, uses a different system that SynthID can't read. Google acknowledges the tool is powerful but doesn't guarantee catch-all coverage.
Who Should Care
- News consumers**: Verify suspicious clips before sharing
- Buyers**: Check product photos before purchasing based on online imagery
- Educators and journalists**: Fact-check visual evidence quickly
- Creators**: Confirm whether submitted work contains undisclosed AI elements
Our Analysis
Google opening SynthID to the public is a strategic move toward becoming the default verification layer for AI-generated content. Rather than waiting for third-party tools to fill the gap, Google is embedding itself as the trusted arbiter for content created through its ecosystem and affiliated platforms.
The Meta gap is the notable weakness. Meta's separate watermarking system creates detection blind spots that users must navigate manually or through Meta's own tools. This fragmentation means no single tool covers the entire AI-generated content landscape yet. Until a universal standard emerges—either through industry consensus or regulatory mandate—verification will require checking multiple systems.
What to watch: Apple's upcoming SynthID integration will expand detection coverage. Watch for other major AI platforms to announce detection partnerships or push for standardized watermarking practices. The real test will be whether detection becomes fast and smooth enough to integrate into everyday browsing habits.
Key takeaways
- SynthID Detector is free and works in any browser—no downloads needed
- It detects watermarked content from OpenAI, NVIDIA, Kakao, and soon Apple
- It cannot detect Meta's Content Seal or unwatermarked AI content
- Use it as one tool in your verification toolkit, not a definitive answer
Frequently asked questions
Does SynthID Detector work on files from any AI generator?
It detects content watermarked with SynthID technology from participating creators. AI content created without SynthID watermarking won't be flagged.
Is the uploaded content stored or shared?
Google states that uploaded files are processed for detection but not stored or shared publicly.
Can I use SynthID Detector on my phone?
Yes, it works through any web browser on desktop and mobile devices.
What if a photo passes the SynthID test—is it definitely real?
Not necessarily. SynthID can confirm AI generation but can't definitively prove something is authentic if no watermark is detected.
Are there alternatives to SynthID Detector?
Yes, tools like the Hive AI Detector browser extension use machine learning to identify AI content, though accuracy varies.