AI images are becoming more difficult to detect. Google believes it has a solution.

Google’s new tool called SynthID is impervious to tampering, making it an important step toward policing the proliferation of bogus photos.

As tech companies strive to develop their AI products, artificial intelligence-generated photographs are getting more difficult to discern from genuine ones. As the 2024 presidential race heats up, there is growing concern that such photographs will be exploited to convey misleading information.

Google launched a new tool called SynthID on Tuesday, claiming that it might be part of the solution. The technology embeds a digital "watermark" straight into the image, which is invisible to the human sight but detectable by a computer that has been trained to read it. Google claims that its new watermarking technology is impervious to tampering, making it an important step toward policing the proliferation of bogus photos and slowing the spread of disinformation.

For the time being, the Google tool is only available to some paying users of its cloud computing service, and it only works with photos created with Google's image generating tool, Imagen. Customers are not required to utilize it, according to the corporation, because it is still in the testing stage. According to Pushmeet Kohli, vice president of research at Google DeepMind, the company's AI unit, the ultimate goal is to help establish a system in which most AI-created photographs can be easily identified using embedded watermarks.