Management agencies and policy makers often use predictive models of how climate is changing to help develop and choose intervention strategies, whether those plans are for climate adaptation strategies, energy policies, or global targets. Increasingly, sets of climate models are being used concurrently to represent the scientific uncertainty in these predictions. But are these climate models working? USGS, The University of Queensland and the US National Snow and Ice Data Centre have developed methods for evaluating climate predictions and detecting when they are failing to capture what is actually being observed. Using their methods they found that observations of summer Arctic sea-ice extent are falling within the bounds of the current set of climate models, but are now favouring those climate models that predict an ice-free Arctic in the summer around 2055. Early detection of failure of a set of climate models can trigger the work needed to diagnose the failure, build better climate models, and ultimately, improve the predictions used as the basis of climate policy and adaptation decisions.