Paste a text that may have come from Meta AI and see which sentences read like output from an AI chat. The demo below checks a section of up to 3,000 characters without signing in; with a free account you see all seven detectors side by side.
Six more detectors
Meta AI is Meta's assistant in WhatsApp, Instagram and Messenger and on Facebook. Meta also releases models with open weights, such as the Llama series, so its technology runs in third-party apps too.
The check does not find out where a text was written. It measures how even the sentences are, the paragraph structure and how predictable the word choice is, which is what the output of large models has in common.
Which model wrote the text. The result does not say that the text came from Meta AI specifically. The differences between models are smaller than the differences caused by the prompt or by later edits, and an independent detector has no key to their watermarks.
Nor can it reliably recognise text that someone thoroughly rewrote, translated or mixed with their own after generating it. With text like that the detectors disagree and the result is borderline.
Text from Meta AI is often short: a message, a photo caption or a comment. The check needs at least 100 characters and with just a few sentences the result is indicative only, so it cannot say anything reliable about a single message.
A longer text, such as a post or an article, tells you more. Delete emoji and hashtags before checking, because they are not part of the style.
A high score is not proof that the text came from Meta AI, and a low one is not proof of the opposite. No AI text detector is one hundred per cent accurate.
When you check someone else's text, take the result as a basis for questions: why the author chose this source, what the term in the second paragraph means, how they reached the conclusion. When you check your own, it shows you the passages worth rewriting in your own words.
Llama is a series of models from Meta that also run outside its apps. The check does not distinguish between them, because it measures the text and not which model wrote it.
Yes, if it has at least 100 characters. With a short message, though, the result is indicative only and proves nothing on its own.
At least 100 characters; a continuous paragraph or a whole chapter is more reliable. With just a few sentences the result is indicative only.
Yes. The check measures the properties of machine writing in general, not the signature of a particular model, so you can check text from ChatGPT and other models in the same way.
Texts up to 300,000 characters, seven detectors at once, colour-marked sentences and a history of your checks. The account is free and needs no card.