MIT Researchers Lead $53 Million ARPA-H Initiative to Revolutionize Weed Control and Protect Public Health Using Generative AI

The Massachusetts Institute of Technology (MIT) is spearheading a groundbreaking, multi-institutional initiative funded by up to $53 million from the federal government to transform how agriculture manages weeds while safeguarding human health and national security. Awarded by the U.S. Advanced Research Projects Agency for Health (ARPA-H)—an agency traditionally focused on radical breakthroughs in human medicine—this major funding allocation underscores the intrinsic link between agricultural inputs, environmental persistence, and public health outcomes.

Operating under ARPA-H’s Systems for Tracking and Resilience in Efficient Agricultural-input Management (STREAM) program, the project is officially titled Precision Learning for On-target Weed control (PLOW). The endeavor brings together top-tier academic institutions and biotechnology pioneers to address a brewing crisis in modern farming: the rapid proliferation of herbicide-resistant weeds, dwindling efficacy of legacy chemicals, and growing concerns over the health impacts of agricultural runoff on the broader population.

The Dual Crisis Facing Modern Agriculture

Every year, farmers across the United States apply more than one billion pounds of herbicides to protect crop yields. These chemical inputs are foundational to modern food production, preventing devastating losses from invasive weeds and supporting regenerative farming practices—such as no-till agriculture—that preserve soil integrity and prevent erosion. However, decades of reliance on a narrow set of chemical formulations have triggered an evolutionary counter-response in target plant populations.

Weeds are rapidly acquiring genetic mutations that render common herbicides ineffective. Prominent agrochemicals like glyphosate (the active ingredient in Roundup) and atrazine are losing their potency against stubborn, resistant strains. Christopher Voigt, head of MIT’s Department of Biological Engineering and the principal investigator for the PLOW project, warns that this biological resistance represents a profound agricultural emergency.

Beyond resistance, supply chain vulnerabilities pose a severe national security threat. The United States remains heavily dependent on imported phosphorus, a critical chemical precursor required to manufacture glyphosate. A disruption in this supply chain, combined with mounting weed resistance, could leave American agriculture without the core tools necessary to maintain high crop yields and affordable food prices.

At the same time, the widespread use of traditional herbicides carries significant public health implications. Trace amounts of these chemicals routinely infiltrate public drinking water supplies and food distribution networks, resulting in low-level, chronic exposure for the general population. Meanwhile, agricultural workers face heightened occupational exposure, which epidemiological studies link to elevated risks of cancer, as well as endocrine and metabolic disorders. The PLOW project aims to resolve this tension by engineering weed-management solutions designed from the ground up to be safe for humans and the environment.

A Multidisciplinary Research Consortium

To tackle a challenge of this magnitude, the PLOW project integrates expertise from biological engineering, artificial intelligence, synthetic biology, and agronomy. The consortium unites researchers from MIT, the Broad Institute of MIT and Harvard, the University of North Carolina at Chapel Hill (UNC), the University of Tennessee at Knoxville (UTK), the University of Illinois at Urbana-Champaign, and agricultural biotechnology firm Robigo.

The initiative is structured into three specialized research teams, each aligned with specific technical milestones outlined by the STREAM program:

1. Generative AI and Computational Design: Tasked with predicting and designing novel molecular candidates, including chemical compounds and small interfering RNA (siRNA), tailored to disrupt vital plant enzymes without harming humans.
2. High-Throughput Screening: Responsible for rapidly testing billions of AI-generated candidate molecules in laboratory settings and feeding screening data back into computational models to refine their predictive accuracy.
3. Greenhouse and Field Evaluation: Charged with assessing how promising candidate molecules perform under real-world agricultural and environmental conditions.

The Technological Approach: Generative AI and RNAi

At the heart of the PLOW project is the deployment of advanced generative artificial intelligence models. Traditionally, discovering a new agrochemical takes over a decade and costs hundreds of millions of dollars, relying heavily on trial-and-error laboratory synthesis. The MIT-led team intends to short-circuit this timeline by leveraging cutting-edge machine learning.

The AI group is led by prominent computational and chemical engineering experts, including 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.

These researchers will design algorithms capable of generating billions of virtual molecular candidates. The models will optimize not just for efficacy—such as targeting EPSP synthase, the crucial plant enzyme inhibited by glyphosate that is necessary for synthesizing essential amino acids—but also for multiple safety and economic parameters. Specifically, the AI will screen for molecules that possess a low risk of inducing resistance mutations, feature low synthesis costs, and present minimal toxicity to humans and non-target ecosystems.

Rather than relying solely on traditional small-molecule chemicals, the project will heavily explore small interfering RNA (siRNA). RNA interference (RNAi) technology allows scientists to silence specific genes within target weed species, offering a highly specific mode of action that degrades naturally in the environment and drastically reduces toxicity risks for mammals.

Once the generative models produce candidate libraries, a high-throughput screening team led by Angela Koehler, the Charles W. and Jennifer C. Johnson Professor of Biological Engineering at MIT, and Jaime Cheah of the Broad Institute will physically test the molecules. According to Voigt, this phase is iterative: screening results will be fed back directly into the computational models, allowing the AI to continuously improve its predictive accuracy.

Rigorous Validation and Environmental Remediation

After narrowing down the candidate pools, the third phase of the project transitions to rigorous biological validation. Greenhouse studies will evaluate how the top-performing molecules perform under varying environmental and agricultural stressors.

This validation team features leading agronomists and synthetic biologists, including Neal Stewart and Scott Lenaghan, co-directors of the Center for Agricultural Synthetic Biology at UTK; Patrick Tranel, the Tom and Peggy Ainsworth Professor of Crop Sciences at the University of Illinois; Tim Willson, Distinguished Professor at the UNC Eshelman School of Pharmacy; and Andee Wallace, CEO and founder of Robigo. The consortium anticipates completing the first greenhouse tests of initial RNAi candidates before the conclusion of the program’s first year.

In addition to developing targeted weed killers, the PLOW project addresses the environmental persistence of existing chemicals. The research team will engineer living microbial systems—utilizing modified bacteria, fungi, and specialized RNA molecules—designed to degrade agrochemicals before they can accumulate in soil, enter the food chain, or leach into municipal water supplies. These engineered microbes will convert hazardous chemical residues and formulation ingredients into harmless biomass, providing a scalable biological safety net for farms and watersheds.

Broader Implications for National Security and Public Health

The convergence of agricultural science and advanced medicine represented by ARPA-H’s backing of the PLOW project highlights a profound shift in federal research strategy. By treating environmental chemical exposure as a critical public health determinant, agencies are increasingly recognizing that chronic disease prevention begins at the source of production.

If successful, the MIT-led initiative could fundamentally decouple American agricultural productivity from reliance on imported chemical precursors and toxic, persistent herbicides. By establishing a paradigm where crop protection agents are designed using predictive AI for human safety and deployed alongside biological degradation safeguards, the PLOW project offers a blueprint for the future of sustainable, secure, and health-conscious global food production.

Funding for this research is provided by the Advanced Research Projects Agency for Health (ARPA-H) under agreement number 140D042690030. The views and conclusions contained herein are those of the authors and researchers, and do not necessarily represent the official policies or endorsements of the United States government.