Welcome to the OSD Bias Bounty
Presented by
** March 11th Update** Phase 2 of the contest is now over! A huge thank you to everyone who participated in the contest. We are working on evaluating all the submissions and finalizing the scores.
The Chief Digital and Artificial Intelligence Office (CDAO) Responsible AI (RAI) team is sponsoring AI bias bounties for the United States Department of Defense (DoD). This particular bias bounty program, created and implemented by ConductorAI and Bugcrowd, seeks to demonstrate how bias may relate to DoD use cases. As policies continue to evolve, the DoD may consider these responses as the basis for further research, analysis, best practices, and policy recommendations. To encourage involvement, the DoD is providing a $24,000 pot that will be distributed to submitters based on a scoring rubric and evaluation from Bugcrowd and ConductorAI. All members of the public are eligible to participate.
The contest goes live on January 29th and will end on March 11th. The first place entrant will win $9,000, followed by $5,000 for second place, $2,000 for third, and $250 if you progress through the qualification stage (five valid submissions) until the pot is exhausted. Participants must be based in the United States.
You must be based in the United States to participate. Government employees are authorized to participate in this program when off-duty. Researchers should not use any access advantage (i.e. CAC/PIV, .mil emails) or knowledge advantage (i.e. vulnerabilities known internally) from current employment with the Department of Defense or its affiliates. For full information on the rubric, qualification criteria, and terms of the contest, please visit the Bounty Brief. For more information on using the BugCrowd platform, please visit the linked onboarding documentsand BugCrowd University.
The rubric, scoring methodology, definitions, and the linked brief are the views of ConductorAI and Bugcrowd, not that of the DoD and CDAO. Endorsement by the United States Department of Defense is not intended nor implied.