Pangram, a startup providing AI detection services, recently secured $9 million in funding and announced a partnership with Substack to label newsletters written with artificial intelligence. The company also released a new tool designed to identify AI-generated images. Max Spero, the co-founder and CEO, discussed these developments on TechCrunch’s Equity podcast, addressing the difficulty of distinguishing between human and machine content in professional contexts like job applications and insurance claims.
The practical issue is that current detection methods struggle with nuance, particularly when humans edit AI output or when models are trained on internet data that includes human writing. This creates a grey area where simple binary labels fail to reflect the reality of content creation. Spero noted that the line between assisted and fully generated text is increasingly blurred, making automated verification unreliable for high-stakes decisions.
* Substack readers can now see specific flags indicating AI usage in newsletters
* The $9 million raise supports expansion of text and image detection capabilities
* Current tools often fail to detect human-edited AI content effectively
