seahorse emoji
In plain wordsA viral meme in which chatbots, asked whether a seahorse emoji exists, insisted it did and then spiraled through wrong emojis trying to produce one — though there was no such emoji.
A failure in which a large language modellarge language model, n.: A computer program trained on enormous amounts of text to predict the words that come next. At vast scale, that simple skill lets it chat, answer questions, translate, summarize and write code., asked whether a seahorse emoji exists, confidently says it does and then, trying to output it, produces a wrong emoji (often a tropical fish or a horse), notices the error and tries again — sometimes looping through many attempts and apologies. No seahorse emoji existed in Unicode at the time, yet many people remember one (a 'Mandela effect'), so the model's belief mirrors a common human false memory in its training data. Interpretability-minded analyses suggest that the model builds an internal representation of 'seahorse emoji' with no matching output tokentoken, n.: The small chunk of text an AI language model reads and writes — often a word, part of a word or a punctuation mark. AI usage, limits and prices are counted in tokens., emits the nearest available one, and then reads its own mistake back from the context. The case became a popular illustration of models' limited self-knowledge and of hallucinationhallucination, n.: When an AI states something false or made up — a fake fact, quotation or source — fluently and confidently, as if it were true. arising from missing rather than wrong information.
Ask it about the seahorse emoji and watch it apologize six times in one answer.
Origin
Scattered earlier posts showed the behavior, but the meme had gone viral by early October 2025, with screenshots of ChatGPT and other chatbots caught in emoji loops spreading on Reddit, X and TikTok. Explanations followed, notably Theia Vogel's logit lenslogit lens, n.: A quick trick for peeking at what a language model is 'thinking' partway through: read its half-finished internal state as if it were the final answer, one layer at a time. analysis of an open-weight model, posted on 4 October 2025. A seahorse emoji was proposed to Unicode but rejected in 2018.
Usage note
Compare the strawberry problemstrawberry problem, n.: The once-famous failure of AI chatbots to count the r's in 'strawberry' correctly — a simple question that exposed how these models see words in chunks rather than letters., another viral failure rooted in how text is represented as tokens.
Related
- strawberry problemstrawberry problem, n.: The once-famous failure of AI chatbots to count the r's in 'strawberry' correctly — a simple question that exposed how these models see words in chunks rather than letters.
- hallucinationhallucination, n.: When an AI states something false or made up — a fake fact, quotation or source — fluently and confidently, as if it were true.
- logit lenslogit lens, n.: A quick trick for peeking at what a language model is 'thinking' partway through: read its half-finished internal state as if it were the final answer, one layer at a time.
- tokenizationtokenization, n.: The first step in feeding text to a language model: chopping it into tokens — whole words, word pieces or characters — and turning each into a number the model can work with.
- jagged intelligencejagged intelligence, n.: The uneven pattern of AI abilities: today's systems can solve very hard problems yet fail at some a child could manage, so their skills form a spiky profile rather than a smooth one.
- calibrationcalibration, n.: How well a model's stated confidence matches reality: when a well-calibrated system says it is 80% sure, it turns out to be right about 80% of the time.
Mentioned in
strawberry problemstrawberry problem, n.: The once-famous failure of AI chatbots to count the r's in 'strawberry' correctly — a simple question that exposed how these models see words in chunks rather than letters.