source: techcrunch ai: google’s deepfake detector system used to debunk mcconnell hoax pic
level: technical
a fake image of senator mitch mcconnell in a hospital bed spread online earlier this week. the picture showed him covered in tubes and looking very ill. it was shared on reddit and x, but fact-checkers at snopes soon proved it was not real. they used google's synthid system to detect an invisible watermark that marked the image as ai-generated.
synthid is a tool that adds a hidden signature to pictures made by some ai models. this signature stays in the image even after screenshots or sharing across platforms. google launched it in 2025, and openai joined the program in may 2026. however, not all ai companies use it. anthropic does not participate. users can check for the watermark by asking a gemini model or using openai's verification tool.
the mcconnell hoax shows how synthid can help fight deepfakes. the senator's health has been a topic of rumors since he went to the hospital on june 14. the fake image tried to exploit that uncertainty. because the watermark survived multiple shares, snopes could quickly confirm the image was fake. this case is a rare public success for the technology, though its reach is limited to tools that choose to add the watermark.
why it matters: it shows a working method to spot ai-made images, which helps reduce misinformation online.
source: techcrunch ai: google’s deepfake detector system used to debunk mcconnell hoax pic