In a post on X (formerly known as Twitter), TestingCatalog claimed that the social media giant was working on a new detection tool for Meta AI. The publication also shared a screenshot of the platform, highlighting the new “AI Detector” menu option.
Meta is working on AI Detector feature for Meta AI.
100% AI 🤖 pic.twitter.com/pdfrsResj1
— TestingCatalog News 🗞 (@testingcatalog) March 15, 2026
While the publication did not specify, the under-development feature was likely surfaced by turning relevant flags on inside the code. However, since the AI Detector is not active from the server side, the page could not be opened. In the TestingCatalog screenshot, tapping on the feature just shows it as a broken link.
Due to limited information, the feature’s full range of functionalities cannot be assessed. There is a possibility that it might just let users see AI-generated or enhanced text, but it might also be able to detect AI images, audio, and video.
Additionally, it is not certain if the AI detection tool can spot AI-generated content from any large language model, or only those developed by Meta. A similar detection tool for Gemini is publicly available, and the social media giant might be following in the same footsteps.
However, since the feature was discovered within the code of Meta AI, and the company is yet to announce it officially, the abovementioned information should be taken with a pinch of salt. Notably, a new report claimed that Meta’s planned new large language models (LLMs), codenamed Avocado, are facing delays again as their performance is not on par with rival models in the market.
