OpenAI Image 2.0 claimed to generate an existing image

OpenAI recently announced Images 2.0 - a new image generation model. In their blog post, under "photorealism" > "College Campus" they give the following example:

The image is claimed to be generated by OpenAI's Images 2.0, however, it is a carbon copy of an image taken by Taryn Chung of Sun File Photo for the cover of a Cornell Sun article: https://www.cornellsun.com/article/2026/03/cornell-accepts-5-776-students-to-class-of-2030.

Did the models weights memorize this image? Did gpt-5.4 pretend to call images 2.0 but just returned an image from a web search? Did the internal tester simply forget to enable the Images 2.0 tool? A HN comment pointed out the image is likely too recent to be in the training data. But I'd love to know if this is a model alignment issue or a simple error.