The recent news of Elon Musk attempting to take over Twitter has put a spotlight on the implications of free speech in the age of social media. The decisions made by these companies as to what content is allowed and what content is moderated are powered by AI-based algorithms called “recommendation systems.” These systems are effective in maximising user engagement, but may come at the expense of undermining the public goods that media companies have traditionally produced for society. In this article, Bullock and Korinek argue that it is crucial to align the objectives programmed into recommendation systems with broader societal values. This would enable social media companies to play a more beneficial role in society, by better balancing the public role of connecting and informing people, and the private role of financing their activities and earning a profit. By doing so, these companies could more effectively navigate the tension between their public and private roles, and ensure that their recommendation systems are in line with the values of our society.