OpenAI's new image watermarks make it easier to spot AI fakes - here's how

["Today, OpenAI announced what it calls content provenance signals across its image ecosystem. In other words, it's tagging its AI-generated images as AI-generated. This is not new, as OpenAI and other AI tools have been embedding metadata in AI-generated images since 2024.

However, the metadata tagging was pretty easy to defeat. What is new is that OpenAI is upping its image ID security game with some fancy new tech.", 'The new technology uses steganography, the practice of embedding cryptographic information in plain sight. According to modern research, this technique has been used for centuries, dating back to around 440 BC.

Herodotus of Halicarnassus wrote about how a message was tattooed on a shaved head and then concealed when the hair regrew. This technique was used as recently as World War II. Steganography has been used in digital images for years to embed text information among the millions of pixels that make up a picture.', 'OpenAI has been embedding metadata in images generated by DALL-E 3, ImageGen, and Sora since 2024.

You can use a tool like Content Credentials to examine that data. However, when a screenshot of an image is taken, which captures the pixels but not the underlying metadata, the metadata is lost. OpenAI and Google are trying to fix this by developing more robust watermarking techniques.', "OpenAI has become a C2PA Conforming Generator Product, which provides a trusted way for platforms to read, preserve, and pass along the provenance information attached to its content.

Additionally, OpenAI is incorporating Google DeepMind's SynthID technology, a multimodal digital watermarking mechanism that embeds invisible digital watermarks in text, images, video, and audio. This technology embeds a subtle steganographic-like signal into images right when they're generated, which can be detected by tools even after resizing, cropping, compression, and color adjustments.", "OpenAI is also announcing the availability of a public verification tool that can be used to see if something was generated by one of OpenAI's AI tools. The tool, which can be accessed at https://openai.com/research/verify/, is expected to be useful in conjunction with SynthID.

According to OpenAI, 'No single provenance technique is enough on its own. We believe a strong approach combines shared standards, durable watermarking signals, and public verification.'", "The introduction of these new technologies aims to contribute to a more interoperable provenance ecosystem. With the increasing use of AI-generated content, the ability to identify and verify the origin of images is becoming increasingly important.