This report surveys the landscape of potential security threats from malicious uses of AI, and
proposes ways to better forecast, prevent, and mitigate these threats. After analyzing the …

The ongoing pandemic of coronavirus disease 2019 (COVID-19), which is caused by severe
acute respiratory syndrome coronavirus 2 (SARS-CoV-2), is a major threat to global health …

Rapid progress in machine learning and artificial intelligence (AI) has brought increasing
attention to the potential impacts of AI technologies on society. In this paper we discuss one …

We introduce four new real-world distribution shift datasets consisting of changes in image
style, image blurriness, geographic location, camera operation, and more. With our new …

We propose a new test to measure a text model's multitask accuracy. The test covers 57 tasks
including elementary mathematics, US history, computer science, law, and more. To attain …

Large language models trained for safety and harmlessness remain susceptible to
adversarial misuse, as evidenced by the prevalence of “jailbreak” attacks on early releases of …

We introduce two challenging datasets that reliably cause machine learning model performance
to substantially degrade. The datasets are collected with a simple adversarial filtration …

… (i, j). In order to speed up computation, for each update, we randomly pick it and only compute
gradients for pairs (it,j),j = it, requiring only 9 top eigenvector computations in each step. …

… “This research study is being conducted by the Steinhardt Group at UC Berkeley. For questions
about this study, please contact Dan Hendrycks at hendrycks@berkeley.edu. In this study…

Abstract Machine learning systems trained on user-provided data are susceptible to data
poisoning attacks, whereby malicious users inject false training data with the aim of corrupting …