In 2022, she published "Governing the Latent Space," a treatise arguing that AI models are simply the world’s largest, messiest hard drive. She launched , a consulting firm that helps companies integrate flagged AI-generated content alongside traditional photography.
In the sprawling digital ecosystems of the 21st century, where data is generated in petabytes and brand identity shifts across hundreds of platforms, the unsung heroes are often the architects of organization. Among those architects, Carola Cott has emerged as a pivotal, if understated, force. While not a household name like Zuckerberg or Musk, within the circles of enterprise content strategy, digital rights management, and global branding, Cott is considered a revolutionary. carola cott
Her current stance is nuanced: AI assets must be labeled with "Provenance Hashes"—a cryptographic signature showing whether an image was born from a camera or a prompt. “A library without provenance is a fiction section,” she recently told The Verge . No industry innovator is without detractors. Carola Cott has faced criticism over her stringent views on "Digital Austerity." She advocates for aggressive data purging—deleting old marketing campaigns within 18 months—which has angered brand historians and archivists. In 2022, she published "Governing the Latent Space,"
Furthermore, her insistence on rigorous metadata entry has been labeled "elitist." Smaller brands argue that her systems, while brilliant, require a level of discipline and budget that only enterprises can afford. Cott’s typical response is blunt: "If you can’t afford to label it, you can’t afford to keep it." Now 54, Carola Cott splits her time between a farmhouse in the Swedish countryside and a research lab in Berlin. She is currently working on a protocol for "Video Deep Indexing," which uses spatial recognition to allow users to search for specific objects inside videos (e.g., "find the scene where the protagonist holds a blue umbrella in the rain"). Among those architects, Carola Cott has emerged as