Social media platforms make difficult decisions about the speech that they permit, and the speech they remove from their platforms. They permit heads of state to propagate ideas that are blocked when expressed by ordinary citizens. They censor activists’ demands for independent states, flagging them as hate speech or terrorist content. Since the social media platforms play a central role in public discourse, their self-regulatory norms matter. Through the architecture of the platforms and powerful algorithms, social media companies control how users engage with each other and with the platform. The architecture enables them to set the terms for engagement, from how users respond to arguments, to how they share photographs, or report abusive content. The algorithms enable fine distinctions of content-moderation, with different content being seen in different jurisdictions, content that cannot be shared at all, and content that will be online but down-ranked so it loses visibility. This power that the social media companies wield over speech online, and therefore over public discourse more broadly, is being recognised as a new form of governance. It is uniquely powerful because the norms favoured by social media companies can be enforced directly through the architecture of the social media platforms. There are no consultations, appeals or avenues for challenge. There is little scope for users to reject a norm enforced in this manner. While a blatantly illegitimate norm may result in uproar, choices made by social media companies to favour existing local norms that violate international human rights norms are common enough. This chapter is about the role of social media platforms. It discusses the role and influence of private companies, especially social media companies, in creating norms affecting freedom of expression. Towards this, it begins with an introduction to the role of the Internet in public discourse, followed by a discussion of how the architecture of the internet enables social media companies to influence speech norms. The paper then discusses the factors that influence the speech norms created by platforms and ends on a cautionary note about how the international human rights norm of freedom of expression may being threatened by the platforms’ self-regulatory norm-setting for online speech.