This competition does not require the use of a dataset. The result of this competition will be a dataset aimed as a benchmark for safety of text-to-image models.