Google DeepMind Introduces Backstory: An Experimental AI Tool for Image Context

Google DeepMind Introduces Backstory: An Experimental AI Tool for Image Context

On July 21, 2025, Google DeepMind announced the introduction of Backstory, an experimental artificial intelligence tool designed to help people explore the context and origin of images encountered online. This initiative builds on a paper published last year that examined how context and provenance can determine trustworthiness, aiming to empower users to make informed decisions about visual content.

Backstory operates by taking an image and a written prompt, then investigating whether the image was AI-generated, where and when it has previously appeared online, and whether it has been digitally altered. The tool quickly provides users with helpful information, responds to follow-up prompts, and describes how the image has been used and how its story may have evolved over time. It also generates easy-to-read reports of its findings.

A conceptual graphic illustrating Google DeepMind’s Backstory tool, featuring digital Polaroid-style photos floating on a dark blue background with pop-up text boxes showing image analysis details such as source, subject identification, and whether the image is AI-generated.

Built using Gemini, Backstory leverages multiple detection technologies to identify whether an image is real or created with generative AI. It then combines this with a broader assessment of the image's context, including its usage across the internet and metadata, to address the user's query.

Article image
Article image

Your browser does not support the video tag.

Article image

Google DeepMind emphasizes that determining whether an image is AI-generated is not the same as understanding its trustworthiness. An image may not be AI-generated but could still be altered or presented out of context, leading to misleading information. Conversely, an AI-generated image might support an authentic, creative, or factual story. Accurately assessing trustworthiness often requires knowledge of how the image was created and a deeper understanding of its surrounding context.

The company stresses the importance of collaboration among industry, civil society, governments, academics, and users to develop and refine tools and programs that maintain the integrity of the information ecosystem. As Backstory is developed, Google DeepMind is working with trusted testers, including content creators and expert information practitioners, to gather feedback and improve the technology throughout the year.

Interested users can explore Backstory and register interest. The announcement also acknowledges the contributions of numerous individuals and teams at Google DeepMind and Google.

Chia sẻ bài viết này