OpenEvidence AI tool spreads among US doctors as accuracy questions grow

In short: A little-known startup called OpenEvidence says nearly two-thirds of US doctors use its AI search tool, but researchers are still debating how reliable it is.

What's going on

OpenEvidence is an AI-powered search engine for doctors. You can think of it like a research librarian that answers questions in plain language, after reading medical journals and treatment guidelines (official advice doctors use). The company says doctors sent more than 42 million questions, also called prompts, to its system last month.

The Miami-based company was founded in 2022 by Daniel Nadler. It has raised nearly $1 billion in the past 18 months and was valued at $15 billion in a September funding round. Unlike many medical software tools, OpenEvidence is supported by advertising, and it reportedly booked $100 million in ad commitments for 2025 and aims for $400 million to $500 million in ad revenue by the end of 2026.

Hospitals are starting to formalize how the tool is used. For example, NYU Langone said it integrated a version of OpenEvidence into its electronic health records, which are the digital files that store patient information. OpenEvidence has also worked with places like Mount Sinai Health System and Memorial Sloan Kettering.

What to watch

Accuracy and over-reliance are the big concerns. A Mayo Clinic study of five sample cases found OpenEvidence gave accurate, evidence-based recommendations, but other research, including a Nature paper, reported it performed worse than general AI systems like Google Gemini and OpenAI GPT on some medical tests. OpenEvidence also left the EU and UK earlier this year, citing the EU AI Act, and it now faces growing competition from OpenAI and Anthropic as they push further into healthcare.

Source: Arstechnica