The extent to which policies are genuinely responsive to public opinion is a key indicator of democratic performance. The media plays an agenda-setting role between the public and the legislators, serving as a mechanism through which policies can be responsive to the public. Existing literature has explored policy responsiveness, the media’s agenda-setting power, and how the media’s effect on the political agenda is contingent on socio-political contextual factors. However, the literature has yet to provide rigorous empirical evidence with chronological precision that policies do respond to media attention. This study examines the driving forces behind the responsiveness of energy development policies to media discourse with a novel methodological approach. Using a machine-learning approach, the author analyses thousands of state-level media reports and legislation featuring hydraulic fracturing (‘fracking’) issues in 15 US states from 2007 to 2017. To model legislation’s responsiveness to media reports, the author identifies topic models for prevalent topics in news articles published in the period leading up to the proposal of a bill. Logistic regression models are estimated with political and socio-economic factors as the predictor variables and whether the bill targets prevalent topics in the news as the dependent variable. The findings suggest that state government ideology, legislators’ partisan affiliations, and unemployment rates predict state-level policy responsiveness to media attention on fracking issues. This study advances our understanding of policymaking’s democratic implications for unconventional energy development and highlights how policymakers can respond strategically to media attention.