Concept
Detecting manipulated or AI-generated content assets at scale, automatically.
Longer Description
Starting with the advent of GANs and progressing to more modern approaches to generative AI, we now have an explosion of AI-generated or manipulated content. We used to rely on the premise of “seeing is believing” but that has since been eroded. A product or product suite built around this idea of detecting manipulation is one that should exist in the world. Whether or not it matters is a different question (see related reading from our team below).
Other thoughts
- It’s very possible that this happens at the creation layer.
- A company built around this must have very strong/defined views around GTM as there are a variety of sectors where this is “nice to have” but not mission critical, and in fact many where it is mission critical may not have enough spend.
- This perhaps curtails into the idea of knowing what assets went into generating a given novel asset as a payback mechanism to creators. We think this is wishful thinking but Shutterstock’s recent advances are interesting to watch.
Comparable Companies
- Snopes
- Variety of on-chain solutions or pixel detection solutions
- The generative media companies themselves
Related Reading