We have been in an odd dialogue with algorithms from the very inception of the internet. They have been trained to spot offensive words, with the goal of allowing civilized conversation while avoiding trolls, spam adverts and hate speech. But, many of our online spaces now moderate content to suit the needs of advertisers. This can mean a lot of people, especially those from marginalized communities, those with alternative or dissident views, or even a-typically creative people, are silenced - and so valuable voices, and conversations could be lost. But humans are very good with language, better than any algorithm developed until now, and we have always found ways to hack around constraints. The latest instrument in this linguistic arms race? Algospeak. Aleks explores the rise of this new form of social media language, discovers how and why black and queer communities are disproportionately silenced by ‘Ad-safe’ algorithms, and finds out that some of the most effective techniques that could allow us to circumvent AI censorship are rooted in the language of people that had to communicate, and mask themselves, with code, long before the digital world existed.