Amid concerns over algorithmic gatekeeping and the power of digital platforms to serve as engines for polarization and disinformation, we examined the performance of algorithmic recommendation systems as news intermediaries by crowdsourcing search results about newsworthy topics. This study offers a crowdsourced audit of the recommendations made by Google, Google News, Facebook, YouTube, and Twitter to ideologically, geographically, and demographically diverse U.S. participants (N = 1,598), examining the extent to which search algorithms on major platforms personalized results and drove traffic to particular kinds of websites. The findings of our cross-platform analysis show that rather than creating filter bubbles, the sorting mechanisms on platforms strongly homogenize exposure to information, at least among its top results. This effect was evident across search terms and platforms. At the same time, each platform prioritizes different types of content, with professionally produced news dominant on some platforms but not others, and politically conservative mainstays like Fox News being particularly recurrent.