Scientists at Kew Gardens have used artificial intelligence to evaluate extinction risk for every known flowering plant species. The finding: according to Kew's State of the World's Plants and Fungi Report, 45% of flowering plants are at risk of dying out.

Why does this matter? Flowering plants make up about 90% of all plant life. They are the base of nearly every food chain on land. They feed us directly—crops are flowering plants. They heal us—most medicines come from plants. When nearly half of them are threatened, the problem is not just about nature. It touches food, medicine, and survival.

Usually, scientists assess endangered species one by one. It is slow work: each species requires expertise, field trips, and years of study. Most plants have never been formally evaluated this way. Kew scientists trained computers to learn from the plants that have been assessed, then used that knowledge to predict risk for all flowering plants. Their analysis was published in March 2024. The computer gives probabilities, not certainty on each species, but when you need to prioritize where to focus limited effort, probabilities are useful.

Here is a troubling detail: we do not even know how many plant species exist. In 2025, scientists named 125 new plant species and 65 new fungi species. That is 125 species we did not know about until this year. A 2023 Kew study found something worse: three out of every four plants that have not yet been named are likely already under threat. This inverts what we usually expect. Finding a new species no longer means it is safe.

The threat is not evenly distributed. Kew identified 32 regions of the world where plants are very diverse but little is known about them scientifically. Many of these regions are in the tropics, where forests are being cleared rapidly. Where we know the least is often where plants are most in danger.

Conservation planners now have a new tool, but it has limits. The AI assessment cannot replace on-the-ground investigation or legal protections. What it can do is point resources toward regions and plant families where risk appears highest. It fills a gap that has existed for decades.

International conservation agreements now aim to prevent human-caused extinctions by 2030. Yet these agreements focus on species we already know about. If most undescribed plants are already at risk—and we are discovering only a few hundred new species each year—then conservation efforts are trying to protect a target that keeps moving.

This research reveals a core problem: we protect what we know, but most species remain unknown. The moment we discover a plant, it may already be dying. The study does not solve that problem, but it makes the problem clear.