MIT will lead an effort of up to $53 million to develop safer herbicides, next-generation alternatives to today's products, using generative AI. The funding comes from ARPA-H, the U.S. Advanced Research Projects Agency for Health, under its STREAM program (Systems for Tracking and Resilience in Efficient Agricultural-input Management). The announcement was made on November 10, 2026.
The project, called PLOW (Precision Learning for On-target Weed control), is led by principal investigator Christopher Voigt, head of MIT's Department of Biological Engineering. It brings together researchers from the Broad Institute of MIT and Harvard, the University of North Carolina at Chapel Hill, the University of Tennessee at Knoxville, the University of Illinois at Urbana-Champaign, and the company Robigo.
Why new herbicides are needed
The motivation is twofold: weeds are developing resistance to the most common herbicides, such as glyphosate and atrazine, and exposure to these products may contribute to a higher risk of cancer and metabolic and endocrine disorders. In the United States, more than 1 billion pounds of herbicide are used every year.
The role of generative AI
The generative AI team will build models able to predict molecules — chemical compounds or small interfering RNA (siRNA) — that disrupt the same plant enzymes targeted by glyphosate and other herbicides. The models are expected to generate billions of candidates, which will then go through high-throughput screening. The group is led by Connor Coley and Regina Barzilay of MIT and Alexander Tropsha of UNC. Screening will be led by Angela Koehler of MIT and Jaime Cheah of the Broad Institute.
Greenhouse tests and microbial allies
A third team will run greenhouse studies to evaluate the molecules under real agricultural conditions, with the first tests of RNAi-based candidates expected by the end of the program's first year. The project will also develop living microbial systems — engineered bacteria and fungi — able to degrade agrochemicals before they reach food, soil, or waterways.
According to MIT, the safer herbicides could also reduce U.S. dependence on imported phosphorus, a critical ingredient in glyphosate. The AI models will also be optimized for low synthesis cost, low risk of inducing resistance, and low risk of harm to humans and the environment.
Image credit: illustrative photo of herbicide application in a field, from the University of Western Australia (UWA), published in the article "Developing and reinventing old herbicides". The image does not depict the PLOW project.
Read also: WorldParticle: uma única IA do MIT simula areia, fluidos, tecidos e proteínas
MIT News; Unite.AI; ARPA-H — STREAM programTransparency: this article was produced with the assistance of artificial intelligence, based on cross-checking public posts and internet sources, verified against the original announcements. Please consult the original sources to confirm the details.
Muse
