Determining whether a disease outbreak stems from a natural spillover event, a laboratory accident, or a deliberate release has long been one of the most technically demanding problems in biosecurity. It requires rapid analysis of enormous volumes of genomic and environmental data to identify microscopic signals buried in biological noise — a task that has historically outpaced available tools.
To close that gap, the Defense Advanced Research Projects Agency (DARPA) launched the Bio-Attribution Challenge in March 2026. On July 1, the agency announced the final prize winners of the competition, concluding a virtual event that distributed a total of $180,000 across two rounds of competition to teams developing next-generation computational attribution tools.
A test designed to simulate real-world conditions
The competition’s final phase, designated the “Determination – Attribution” phase, pushed participants to process datasets ranging from 600 to 800 terabytes within a strict 24-hour window — a scale and time constraint designed to reflect the operational demands of an actual biological incident response. Teams were required to combine deep genomic sequence screening with geospatial and temporal metadata analysis, and were given minimal prior information about the samples to maximize realism. All challenge data were curated by Lawrence Livermore National Laboratory to simulate credible attribution scenarios without the use of actual pathogens or the disclosure of sensitive government information.
Eight teams completed Round 2 and were recognized during a virtual awards ceremony. “The ability to rapidly and accurately identify, characterize, and determine the source of a biological sequence, whether natural or engineered, within an enormously complex environmental sample background is a critical national security capability and U.S. government priority,” said Cmdr. Cody Thornton, DARPA’s deputy program manager for the Bio-Attribution Challenge. Thornton added that the top-performing teams demonstrated methodologies offering “orders of magnitude improvements over current tools.”
Winners across performance categories
The $90,000 distributed in Round 2 brings the total prize pool to $180,000 across both rounds. Round 1 winners were announced in May 2026. The final awards are as follows:
Overall Winners:
Team Crits-Christoph & Hakim – First Place $50,000
Team Kannadasan – Second Place $30,000
Aclid – Third Place $10,000
Best in Show (best overall performance across both rounds):
Team Crits-Christoph & Hakim
Fastest Analysis (fastest methods for executing data analysis):
TwentyTwo Bio (from Round 1)
Most Innovative Method: (most novel mathematical or computational approach):
Team Sylph (Carnegie Mellon University Yu Lab)
Most Data-Efficient: (smallest software package to achieve attribution):
Team Kannadasan
Highest Precision (closest measurements to actual true values):
Pastor21
Highest Accuracy (most consistency and reproducibility in measurements):
Treangen Lab
Why this matters for health security
The inability to rapidly and reliably attribute a biological event to its source (whether natural, accidental, or deliberate) is a recognized vulnerability in global health security infrastructure. Without that capability, public health and defense responses risk being misdirected or delayed. The tools demonstrated through this challenge represent a potential step toward closing that gap, offering faster, more precise computational methods that could eventually be integrated into federal biosurveillance and incident response systems. The competition’s structure also establishes a model for advancing sensitive national security capabilities without exposing classified information or requiring access to dangerous biological materials.
Sources and Further Information:
Bio-Attribution Challenge yields tools to define biothreat origins at speed, scale – DARPA