ChatGPT users have reported the AI bot increasingly inserting Arabic words into its English responses, as seen in viral social media posts. One Reddit user claimed the bot did it "twice on my phone and once on my work laptop," including a recipe list with an ingredient randomly listed in Arabic. In a viral X post, a user recalled the chatbot inserting Arabic while helping write a prompt for a logo. When asked about the gaffe, the large language model claimed it "slipped in by mistake," per a screenshot. Many Reddit commenters reported similar glitches with responses in Armenian, Hebrew, Spanish, Chinese, and Russian. More astute users observed that the issue relates to how the AI system is programmed using tokens, a cybernetic shorthand. As AI bots are trained on many languages, they might insert a foreign word that is shorter and easier to process because it saves on tokens. One Redditor explained that the Arabic word in a recipe post means "low," translating to "low-fat yogurt." In a post about HIPAA, the Arabic phrase translated to "within the USA" so it "did make sense."