The use of new data sources to model humans' behavioural responses to climate change (see P. Palmer and M. Smith Nature 512, 365–366; 2014) raises methodological and ethical issues.

The authors do not mention the importance of call-detail records (CDRs), normally collected by mobile-phone operators for commercial purposes. Compared with data collected by smartphones from global-positioning systems, CDRs have lower location accuracy and differ in their potential for modelling and privacy risks.

The poorest communities in low- and middle-income settings should be at the centre of modelling efforts because they are among the most vulnerable to climate change.

These populations are unlikely to use satellite navigation and social media, and they might not conform to human-mobility models derived from commuting patterns in wealthy countries. In these regions, CDRs can provide important insight — if appropriate privacy protections are in place (see, for example, A. Wesolowski et al. Science 338, 267–270; 2012).

The rights of the individual to control their private data and the needs of researchers and policy-makers to access data for societal good create tensions that are central to the effective modelling of human behaviour (see, for example, Y. de Montjoye et al. Sci. Rep. 3, 1376; 2013). These call for new regulatory and institutional review board processes.