Google’s deepfake detector system used to debunk McConnell hoax pic

Google's innovative SynthID technology has successfully exposed a viral deepfake image that misrepresented Kentucky Senator Mitch McConnell in a hospital setting. The image, which depicted McConnell in distress and connected to various medical tubes, quickly gained traction on social media platforms like Reddit and X. However, it was swiftly refuted by the fact-checking website Snopes, which identified the presence of SynthID's unique watermark embedded in the image. This incident marks a notable success for SynthID, a tool designed to combat misinformation generated by AI. The watermark, intended to be invisible to the naked eye, enabled the verification of the image's authenticity, confirming it as artificially created. The scandal surrounding McConnell's health has intensified since his hospitalization following an emergency call on June 14, leading to rampant speculation about his well-being. Yet, in this case, the evidence of distress was proven entirely fabricated. SynthID, which debuted at Google's I/O developer conference in 2025, embeds an invisible signature within AI-generated images. This signature remains intact even when images are captured or shared across various platforms, as was the case with the McConnell photo. Notably, the technology relies on the cooperation of image-generation tools, with Google's Gemini models incorporating the watermark since the program's inception. OpenAI also joined the initiative in May 2026, while Anthropic has not yet participated. Users can verify the authenticity of images by querying a Gemini model or utilizing OpenAI's public verification tool.