Following decades of experimental investigations involving individual participants, large-scale open data and computational advances newly enable the study of how social attitudes have changed over the long term, at the level of societies. In this project, we harness data from the Project Implicit International Dataset (Charlesworth et al., 2023), collected continuously between 2009 and 2019 from 1.4 million+ participants across 33 countries, to examine global trends in explicit (self-reported) and implicit (automatically revealed) attitudes toward five social group targets (age, body weight, sexuality, skin tone, and race). Bayesian time-series modeling using Integrated Nested Laplace Approximations (INLA) revealed changes toward less bias in all five explicit attitudes, with effect sizes ranging from 18% for body weight to 43% for sexuality. Implicit attitudes also decreased in bias by 36% for sexuality; implicit age, body weight, and race attitudes remained constant; finally, and uniquely, implicit skin tone attitudes increased in bias by 20%, with the pace of change accelerating since 2016. Results were robust across changes in sample composition and unlikely to be explained by age or cohort effects. Consistent patterns of change across all explicit attitudes toward less bias imply the presence of increasingly widespread norms against the expression of negativity toward stigmatized groups. The findings involving implicit attitudes underscore the potential of these attitudes to exhibit meaningful, long-term change toward either less or more bias. Moreover, the variability in implicit attitude trajectories suggests that the relevant sources of change are likely to be macro-level events affecting a particular attitude domain.