A $100,000 prize pool anchors the new ARC White-Box Estimation Challenge hosted by AIcrowd. Participants must improve estimation algorithms for random MLPs based on recent research. The competition starts this week with a warm-up round. This effort targets technical interpretability, providing a concrete benchmark for researchers attempting to decode the internal logic of neural networks.