Text style can reveal sensitive attributes of the author (e.g. race or age)\nto the reader, which can, in turn, lead to privacy violations and bias in both\nhuman and algorithmic decisions based on text. For example, the style of\nwriting in job applications might reveal protected attributes of the candidate\nwhich could lead to bias in hiring decisions, regardless of whether hiring\ndecisions are made algorithmically or by humans. We propose a VAE-based\nframework that obfuscates stylistic features of human-generated text through\nstyle transfer by automatically re-writing the text itself. Our framework\noperationalizes the notion of obfuscated style in a flexible way that enables\ntwo distinct notions of obfuscated style: (1) a minimal notion that effectively\nintersects the various styles seen in training, and (2) a maximal notion that\nseeks to obfuscate by adding stylistic features of all sensitive attributes to\ntext, in effect, computing a union of styles. Our style-obfuscation framework\ncan be used for multiple purposes, however, we demonstrate its effectiveness in\nimproving the fairness of downstream classifiers. We also conduct a\ncomprehensive study on style pooling's effect on fluency, semantic consistency,\nand attribute removal from text, in two and three domain style obfuscation.\n