Google’s deepfake detection system was used to expose McConnell’s hoax image

Google Synth ID The system was used to expose a high-profile hoax image generated by artificial intelligence, in a rare but significant win for the system.

Earlier this week, a photo circulated online that appeared to show Kentucky Sen. Mitch McConnell covered in tubes in a hospital bed in extreme distress. The photo was shared widely on Reddit and Xbut by Wednesday, the venerable fact-checking site Snopes He debunked the image, noting that when scanned, the image is registered as containing the SynthID watermark designed by Google to recognize AI-generated images.

In short, the watermark worked exactly as it was supposed to in the anti-deepfake technology win.

Senator McConnell’s health has been the subject of intense speculation since his hospitalization After an emergency call on June 14. Since then, he has largely disappeared from the public eye, sparking speculation that his health may be deteriorating. But in this case, the evidence was proven to be completely fake.

I set off At the Google I/O developer conference in 2025, SynthID acts as an invisible signature, visible to SynthID algorithms but designed to be unnoticeable to the casual observer. Because the signature is embedded in the image itself, it remains there even when the image is screen-captured across multiple platforms, as McConnell’s image was.

The main limitation of SynthID is that it can only be used when the image generator is actively participating in the software. Gemini models have included the watermark since the program launched in 2025. OpenAI joined in May 2026, as part of A broader effort to combat harmful image generation. Anthropic does not participate in the program.

Users can check if photos contain the watermark by asking or uploading to the Gemini form OpenAI public image verification tool.