Google DeepMind’s Backstory streamlines image fact-checking at India Today
The experimental tool can flag AI fakes, trace where an image has appeared online and bundle the evidence into a cited report, though reporters still verify the findings themselves.
Google DeepMind’s experimental Backstory tool is helping India Today’s fact-checkers sort through misleading images circulating online. It can assess whether a picture appears to be AI-generated or manipulated, and it can trace where the same image has appeared elsewhere on the web.
According to Nieman Lab, a user uploads an image and enters a basic question, after which Backstory decides which authentication tools to run. Built on the Gemini family of large language models, it uses AI agents to choose the sequence of checks, then produces an AI-generated report with citations and a log of the steps it took. For now, the tool is described as free to use.
At India Today, Backstory is mainly used as a first pass, speeding up the opening stage of the verification workflow. Bal Krishna said the team uses it as a quick reference because it gives context and direction, and that it does in one place the work of five different tools and five separate logins. He also called it the first robust tool from a big tech company offered to fact-checkers for image verification.
The tool’s usefulness, in Krishna’s view, is that it can do in one sitting what used to take several steps. Reporters previously might have had to run an image through an AI detector, carry out a reverse image search and piece together a timeline of earlier appearances. Krishna said Backstory does those tasks at once.
Backstory also checks for provenance signals. It can look for SynthID watermarks in images created with Nano Banana, and it can examine C2PA Content Credentials, the provenance system designed to record how digital content was created and modified. Google DeepMind says the tool was developed with trusted testers, including content creators and expert information practitioners, as it gathered feedback through the year.
The limits matter as much as the gains. Backstory carries a disclaimer saying it can make mistakes, and India Today does not cite its AI-generated reports directly in published fact-checks. Reporters are expected to confirm anything they take from the tool on their own.
The system also has technical constraints. It currently processes images only, and its context tracing is limited to images indexed by search engines. Even when Backstory says it has found an image’s origin, the image may have had a longer life outside what the tool can see.
That uncertainty showed up in testing. The Nieman Lab report said Backstory briefly mistook the author’s headshot for another journalist’s image before correcting itself after the author supplied the name. The result is a tool that can accelerate image verification, but still leaves the final judgment with the fact-checker.