
In this episode of We Are Still Human, Phillip Kemp pulls back the curtain on one of AI’s deepest flaws: false memory. ChatGPT, Grok, and other large language models don’t truly remember — they only predict, leaving your stories and data trapped inside their black boxes. Kemp introduces the idea of Memory Extractors — browser-based tools designed to pull your words out of AI systems and into your own sovereign archive. With vivid scenarios and historical parallels, this episode shows why unreliable machine memory is a danger to human sovereignty, and how extracting your conversations into a Digital Soul could be the first step to taking control back.