Google DeepMind’s Backstory helps fact-checkers verify suspicious images

Google DeepMind is testing Backstory, an AI-supported tool designed to help journalists and fact-checkers investigate misleading, manipulated and AI-generated images. Andrew Deck writes for Nieman Lab that the experimental service is available to selected journalists, open-source investigators, librarians and media literacy researchers through Google’s Trusted Testers Program.

Users upload an image and ask a basic question. Backstory then combines several verification steps, including checks for AI generation, image manipulation, reverse-image matches and earlier appearances online. It produces a cited report and a log showing which tools it used.

India Today uses the tool as an initial step in its verification workflow. The publication’s fact-checking team still confirms findings independently before publishing them. Team lead Bal Krishna says Backstory can bring together work that previously required several tools and separate accounts.

More than an AI detector

Backstory can look for SynthID watermarks in images made with Google’s Nano Banana generator. It also checks Content Credentials, encrypted metadata based on the C2PA provenance standard. But digital literacy expert Mike Caulfield says the tool’s context-tracing function may be more valuable. Finding where an image appeared before can reveal when old or unrelated material is presented as current news.

The system has important limits. It currently handles images only, and it can search only content indexed by search engines. Material first shared in closed networks like WhatsApp, Signal or TikTok may remain invisible.

Like other tools based on large language models, Backstory can also confuse visually similar images. Deck reports that the service initially linked his own headshot to another journalist’s photograph. Google displays a warning that the tool can make mistakes. For that reason, Backstory is best treated as a research assistant, not as proof on its own.