AivexaNewsSearch
AI news for builders and product teamsChecked every hour

MIT-led team receives funding to develop next-generation alternatives to modern weed killers

Collected Nov 10, 2026

MIT researchers are leading a program worth up to $53 million, awarded by the U.S. Advanced Research Projects Agency for Health (ARPA-H) through its STREAM program (Systems for Tracking and Resilience in Efficient Agricultural-input Management). The project is called Precision Learning for On-target Weed control (PLOW), and Christopher Voigt, head of MIT's Department of Biological Engineering, is the principal investigator. The work targets next-generation weed killers intended to keep farms productive while reducing risks to human health.

The scale of the problem is large: more than 1 billion pounds of herbicide are used in the United States each year. Herbicides limit crop losses to weeds and support regenerative practices that protect soil, but resistance is eroding their effectiveness, and exposure to some of these chemicals may raise the risk of cancer, metabolic disorders, endocrine disorders and other health problems. Farm workers face the heaviest exposure, yet much of the U.S. population takes in small amounts through food and drinking water.

Voigt frames the issue as a convergence of two pressures. Resistance mutations are spreading among weeds targeted by glyphosate and atrazine, and the supply chains for herbicide precursors raise national security concerns. In his words, the country could find itself without core agrochemicals that underpin current yields and affordable prices. The team also notes that alternatives could reduce U.S. dependence on imported phosphorus, a critical ingredient in glyphosate.

PLOW is organized into three teams matching the STREAM technical areas. The first, the generative AI group, will build models that propose molecules capable of disrupting the same plant enzymes that glyphosate and other herbicides attack. Glyphosate works by interfering with EPSP synthase, an enzyme plants need to make several essential amino acids. The candidates can be either chemical compounds or small interfering RNA (siRNA), which silences gene expression rather than acting as a conventional chemical. The models will be tuned not just for potency but for low synthesis cost, low likelihood of triggering resistance, and low risk to people and ecosystems.

That group is led by Connor Coley, the Warren K. Lewis Associate Professor of Chemical Engineering at MIT; Regina Barzilay, the School of Engineering Distinguished Professor for AI and Health at MIT; and Alexander Tropsha, the K.H. Lee Distinguished Professor in the UNC Eshelman School of Pharmacy. The models are expected to generate billions of candidate molecules, far more than any lab could test by hand, which is where the second team comes in.

The high-throughput screening group, led by Angela Koehler, the Charles W. and Jennifer C. Johnson Professor of Biological Engineering at MIT, and Jaime Cheah, associate director of scientific outreach and collaborative screening at the Broad Institute's Center for the Development of Therapeutics, will test those candidate libraries. Voigt describes a two-way loop: screening identifies molecules that hit their target, and the resulting data flows back to the computational models to improve them. In practice this is an active-learning cycle, a pattern familiar from AI-driven drug discovery, where generated molecules are assayed and the results retrain the generator.

The third team takes the survivors into greenhouse trials to see how each molecule behaves under varied agricultural conditions. It is led by Neal Stewart and Scott Lenaghan, co-directors of the Center for Agricultural Synthetic Biology at the University of Tennessee at Knoxville; Patrick Tranel, the Tom and Peggy Ainsworth Professor of Crop Sciences at the University of Illinois at Urbana-Champaign; Tim Willson, the Harold Kohn Distinguished Professor at the UNC Eshelman School of Pharmacy; and Andee Wallace, a 2020 MIT PhD and CEO and founder of Robigo. The first greenhouse tests of initial RNAi candidates are expected to be finished by the end of the program's first year.

A separate strand of the project involves living microbial systems designed to break down agrochemicals before they reach food, soil or waterways. Engineered bacteria, fungi and srRNA would convert chemicals and formulation ingredients into biomass, an approach the team describes as a scalable way to cut environmental and human exposure. This suggests a containment strategy layered on top of substitution: even as new active ingredients are developed, engineered organisms could reduce residues from existing ones.

Other participants are expected to come from the Broad Institute of MIT and Harvard, UNC-Chapel Hill, UT Knoxville, the University of Illinois at Urbana-Champaign and Robigo. The funding falls under ARPA-H agreement number 140D042690030, and MIT notes that the views expressed are the authors' and not official U.S. government policy.

For developers and product teams working on molecular machine learning, the project is a notable vote of confidence in generative chemistry outside pharma. The stated objectives — modeling resistance risk, synthesis cost and toxicity alongside target binding — point toward multi-objective generation rather than single-score optimization, a design problem that also shows up in materials and enzyme engineering. A likely trade-off is that siRNA and other biological approaches may face different regulatory and formulation hurdles than conventional small-molecule herbicides, and the source gives no timeline beyond the first-year greenhouse milestone.

Why it matters: a $53 million federal program puts AI-driven molecule design at the center of agricultural chemistry, an area where the main incumbent products face resistance and health scrutiny. Teams building generative models, screening pipelines or engineered microbes gain a prominent public-sector reference case, and the closed loop from prediction to assay to retraining is the blueprint likely to spread. Whether the program yields viable field-ready alternatives is an open question, but the first RNAi greenhouse results are due within a year, giving an early signal of how fast this approach can move.

Read at MIT News · AI

Based on reporting from the original publisher. Visit the source for full context and later updates.

Publisher excerpt

With up to $53 million from ARPA-H, the researchers will use AI to generate safer herbicides that reduce dependence on imports and improve the security of the U.S. food supply.