As artificial intelligence becomes increasingly capable of writing like a human, a new question is emerging: can we still tell the difference? What once seemed obvious is now becoming remarkably difficult, as modern language models learn to imitate not only facts and grammar, but also style, tone, and persuasion. In this episode, we explore the evolving language of AI-generated writing, examining how large language models have changed, why their telltale signatures are becoming harder to spot, and what researchers have discovered about the subtle patterns that still distinguish machine-generated prose from human expression. We investigate the ongoing race between increasingly sophisticated AI systems and those trying to detect them, while considering what this means for journalism, education, publishing, and online communication. The story reveal ...