Stanford researchers unveil FYI for transparent AI fact-checking

Researchers from Stanford University’s Computer Science department have released FYI, an open-source browser extension designed to bridge the gap between automated AI fact-checking and manual data verification. Published as arXiv:2608.06804 and set to appear at IEEE VIS 2026, FYI embeds fact-checking directly into the reading environment of data-driven articles. Unlike traditional systems that isolate AI verification from manual inspection, FYI integrates four complementary tools—ranging from fully automated checks to user-driven data exploration—allowing readers to seamlessly transition between AI assistance and their own analysis.

In an exploratory study with 22 participants, the team identified three distinct workflow archetypes: AI-first with manual confirmation, manual-first with AI supplementation, and parallel co-review. Visualization emerged as the primary mechanism for validating AI outputs, with trust in AI shifting dynamically based on consistency and convergence across tools. The findings challenge the notion of AI as a definitive authority, instead positioning it as a starting point for human verification. FYI is available as open-source software for further research at the provided link.

The research highlights the need for flexible, user-driven workflows in AI-assisted fact-checking and underscores the critical role of visualization in building trust. By releasing FYI as open-source, the team aims to spur further innovation in transparent, human-in-the-loop verification systems.