Theia Vogel digs into the phenomenon of why large language models seem to think that there is a seahorse emoji. Spoiler alert: There isn’t one, but GPT-5, Claude, and others are convinced there is — along with many humans:

Maybe LLMs believe a seahorse emoji exists because so many humans in the training data do. Or maybe it’s a convergent belief - given how many other aquatic animals are in Unicode, it’s reasonable for both humans and LLMs to assume (generalize, even) that such a delightful animal is as well. A seahorse emoji was even formally proposed at one point, but was rejected in 2018.

Sure enough, when I prompt GPT-5 asking for a seahorse emoji, I get this reply:

Hilarious. This is the 2025 version of the “How many r’s are in strawberry” prompt that would trip up most models until recently. It shows how LLMs are pattern learning and matching machines and can generate what seems like it ought to exist rather than what actually exists — similar to the Mandela effect experienced by large group of humans that share a false memory.