andrea binhigh stakes designAdversarial attacks are only part of the problem, so we also audited the hardware of this AI systemGo ahead. Bring on the complexity.May 31, 2023May 31, 2023
andrea binhigh stakes designIs your AI a “Clever Hans”?What a counting horse can teach us about evaluating AI tools & systems.Feb 23, 2023Feb 23, 2023
andrea binhigh stakes designAlice, Bob & Eve, the eavesdropperA comic about encryptionJan 14, 2023Jan 14, 2023
andrea binhigh stakes designWe audited RoBERTa, an open source large language modelWe found two security vulnerabilities and a backdoor.Dec 19, 2022Dec 19, 2022
andrea binhigh stakes designDo deepfakes discriminate?Here’s what we learned from bias testing two deepfake detector models.Dec 19, 2022Dec 19, 2022
andrea binhigh stakes designAn “ethical matrix” for FakeFinderThis thought exercise helped us identify the risks of deploying a deepfake detection tool.Dec 16, 2022Dec 16, 2022
andrea binhigh stakes designWhat happened when we audited a deepfake detection tool called FakeFinderTLDR: This AI tool doesn’t do quite what we thought…Dec 14, 2022Dec 14, 2022
andrea binhigh stakes designNew Paper on medRxiv about Evaluating COVID-19 ForecastsHonored to be part of this amazing team of people working on COVID-19 forecasting.Feb 17, 2021Feb 17, 2021
andrea binhigh stakes designVisualize Influenza & COVID-19 Forecasts with VizifluViziflu is an open source tool for visualizing infectious disease forecasts.Jul 3, 2020Jul 3, 2020