New · See every ThinkChat feature in action. Take a look →
ThinkChat
Start free trial

Knowledge AI

Is that photo AI-made? How to use Google’s SynthID Detector, and why a clean result isn’t proof

AI 4 min read

On 7 October 2026, Google opened its SynthID Detector to everyone. It’s a website where you upload an image, video or audio file and it checks for an invisible watermark that some AI tools add to what they create. It’s available globally, in English.

For a small business, this matters more than it sounds. Payment slips, product photos, “proof” screenshots and voice notes now arrive on WhatsApp every day, and AI makes all of them easier to fake. So here’s what the detector can and can’t tell you.

What Google launched

  • A public checker. Google first offered an early version of the SynthID Detector to journalists and media professionals in 2025. Now anyone can use it at synthid.com.
  • Images, video and audio. For images it says whether a watermark is present. For video and audio it points to the parts that carry the watermark.
  • More than Google’s AI. It checks for watermarks from Google’s own AI and from partners that use SynthID: OpenAI, NVIDIA and Kakao, with Apple “soon”, Google says.
  • Already in Gemini, Search and Chrome. You can also upload a file to the Gemini app and ask if it was made or edited with Google AI.

Ars Technica adds two practical details. You need to sign in with a Google, OpenAI or Apple account, and Google told Ars that each user gets “a daily usage quota of approximately 10 image, video, and audio checks.” Google engineers told Ars the limit is there so people can’t use repeated checks to build tools that remove the watermark. Ars adds that you may be cut off sooner if you upload many very similar files.

How the watermark works

SynthID adds a hidden signal the moment AI creates the content: into the pixels of an image or video, or into the sound of an audio clip. People can’t see or hear it. Google says it is designed to survive common changes like cropping, filters, colour changes and compression.

This is different from general “AI detectors” that guess whether something looks AI-made. SynthID looks for a signal that was deliberately put there.

Why a clean result isn’t proof

Google is clear that this is not a general AI detector. A “not detected” result can happen when:

  • The AI tool doesn’t use SynthID. Many AI tools don’t, so the detector can’t see their content.
  • The file was heavily changed. Heavy editing or compression can weaken the watermark. Google says someone who tries many changes may find one that isn’t detected.

Google also says that, in very rare cases, the detector can wrongly flag content that has no watermark, and that a “detected” result means the file was likely made or edited with a SynthID-enabled AI tool. It doesn’t tell you whether AI made the whole file or only edited it.

Google’s own advice is to collect more than one piece of evidence before deciding.

What this means for a small business in Malaysia

1. Use it as one check, not the final word. If a photo or video looks suspicious, a “detected” result is a strong warning sign. A “not detected” result doesn’t clear it.

2. Upload the best copy you have. Google recommends the highest-quality file possible. WhatsApp says media files are usually compressed when shared, so a photo or video sent through a chat may be lower quality than the original. That can make the watermark harder to find, so ask for the original file if you can.

3. Keep your usual checks. For payments, check your bank app, not the slip. For a request that sounds urgent, call back on a number you already have. These checks don’t rely on spotting AI, so they still help when a detector can’t.

4. Be open about your own AI content. If you use AI to make product images or replies, saying so builds trust. Customers increasingly have tools to check.

Where ThinkChat fits

ThinkChat is an AI Chat Agent and multi-channel AI customer service platform. WhatsApp is the main channel in Malaysia, and ThinkChat also works on Instagram, Facebook Messenger, Telegram, website chat and more. It is built to answer customers from information you approved, and hand over to a person when a customer needs one, such as when something needs a human check.

Soft next step

Want an AI Chat Agent to answer your customers 24/7? Talk to Lyia on WhatsApp: +60 11-5515 9317.

Sources

Checked 28 October 2026.

← All articles