Since no conclusive evidence for the nature of physics beyond the Standard Model has yet been found, this motivates adding new analysis methods to our arsenal. Traditional new-physics searches face an inherent trade-off between coverage and sensitivity: enhancing sensitivity occurs through model-dependent cuts, which limit the coverage of the search. Anomaly detection offers a way to break this trade-off by searching for regions which do not match the background distribution without assuming a specific signal model. Over the past decade, such methods have matured from theoretical proposals to fully-fledged analyses applied to real LHC data by both ATLAS and CMS. In this talk, I will review the key developments that have brought the field to this point and highlight results which demonstrate that anomaly detection is both a robust and sensitive probe of new physics.