In response to the human advances, the AI re-trains to explicitly defeat the detection tools. The cycle of GAN begins. Academic competitions appear, but the air is uneasy as researchers question if posting their ideas only helps the bots get stronger.
Well-funded by governments and companies, the bots have a weakness: they are competing entities in shadow wars of politics and commerce, while the humans are collectively pooling their resources. A general consensus emerges: the growth of the detectors must consistently outpace the bots. Aware of their collectivist edge, the humans rapidly consolidate their detection technologies.
Fueled by venture and government investment, the detection companies become export restricted, as Congress considers the models to be critical to national security. Academic competitions become less common as large organizations hide their architectures from the bots. The humans are safer, but eventually lose the open collaborations that gave them their edge. The humans look at each other through the goggles they built. Have they won?
