Can You Tell If That Photo’s AI? Google’s New Tool Can

Google made several new announcements at the I/O 2025 event. A handful of these are related to artificial intelligence features and the Gemini app, which now has 400 million monthly active users. At the same annual developer event, the search titan has also introduced a new tool called SynthID Detector, which is essentially a verification portal to help you identify AI-generated content.

The SynthID Detector tool can identify AI content generated using Google’s tools

In a blog post, Google has explained that its new SynthID Detector verification portal is capable of identifying images, audio tracks, videos, and text created using its own artificial intelligence-powered tools. When you upload a photo, audio track, video, or a piece of text on the portal, it will scan the media for a SynthID watermark. When the SynthID Detector verification portal detects something, it highlights the specific portions of the content that most likely have a watermark.

For audio files, the tool can pinpoint the specific segments where it detects a SynthID watermark. As for photos, it highlights areas where a watermark is most likely present. For those unaware, in 2023, Google launched the SynthID Text tool to help people determine if AI tools like ChatGPT produced a research paper.

The SynthID Detector verification portal will be available on the web, allowing you to upload photos, videos, audio, and text for scanning. Google is starting to roll this tool out to early testers right away, before making it available more widely. The company noted that journalists, media professionals, and researchers around the globe can join the waitlist to gain access to the SynthID Detector.

It appears that Google’s SynthID Detector tool is a response to the growing prevalence of AI-generated content online. According to a report by The Times, deepfake videos have seen a staggering increase of 550 percent from 2019 to 2024. Furthermore, last year, AI created the top 20 most-viewed posts on the social networking platform Facebook in the US. It highlights the growing impact of this technology.

Google designed SynthID to watermark content generated by its AI models, but the company hopes other AI firms will adopt this watermark for their AI outputs. It’s worth mentioning that Google is already offering an open-source version of SynthID, which is available for watermarking text.

Google is partnering with NVIDIA to watermark content generated by the NVIDIA Cosmos model. It’s worth adding that SynthID Detector won’t be the only tool that can identify Google’s watermark. Google says GetReal Security will also be able to verify if a media file contains SynthID.

That said, despite its potential, Google’s SynthID Detector verification portal comes with certain limitations. As of now, it is only capable of identifying content generated using Google’s AI tools. Google stated that this technology is not infallible and some can bypass it, particularly when it comes to text-related content.