Yes. Grammarly's Authorship feature tracks how a document was written and flags content that came from AI tools. Here is how it works, what it can and cannot see, and what to do about it.
How Grammarly handles AI detection
Authorship runs through the Grammarly browser extension in Google Docs, the Grammarly desktop app in Microsoft Word, and the Grammarly editor. As you type, it tags the origin of every chunk of text: human typing, a single paste, an AI suggestion accepted from GrammarlyGO, or AI output pasted in from a third-party tool like ChatGPT.The data produces an Authorship report the writer can export as a PDF and share with an instructor or manager.This is fundamentally different from post-hoc AI detectors like GPTZero, Turnitin, or Copyleaks. Those tools analyze a finished document. Authorship watches the writing process itself, which makes it reliable for detecting paste events and less sensitive to the stylistic false positives that trip statistical detectors, including on non-native English writing.
Accuracy in practice
On direct paste detection, Authorship is essentially deterministic: it either has paste telemetry or it does not. That is high reliability by construction, not by classifier skill. Grammarly publishes no false-positive rate for paste tracking because there is nothing to be false-positive about when the data is a real keystroke log.The trade-off is coverage. Authorship only sees writing done while tracking is on, in places such as Google Docs, Microsoft Word or the Grammarly editor. A student who drafted somewhere else, or in Docs before switching tracking on, then pasted the draft in for a cleanup pass, produces a report that shows almost no typing, because the bulk of the writing happened before Authorship was watching. That can look suspicious to an instructor who does not understand the product.Grammarly's Authorship guide is clear on the limit: Authorship documents the writing process it tracked, and text it cannot trace to a source is labeled unknown.
What to do about it
If a course or employer requires Authorship, the practical workflow is to turn it on where you write, in Google Docs, Microsoft Word or the Grammarly editor, and keep it on from the first draft forward. Pasting research quotes is fine: Authorship logs those as pasted text, and that is accurate. The pattern that raises questions is a long essay that arrives as a single paste with minimal typing after.If AI assistance is permitted and you want a clean Authorship report, letting GrammarlyGO do the AI suggestion rather than pasting ChatGPT output keeps the report clear, since that content is labeled as Grammarly-assisted. For reviewing final text before submission, a free statistical detector such as Leap's AI score checker complements Authorship by checking writing signals that paste tracking does not measure. See our primer on how AI detectors work for the difference between process-based and classifier-based detection.
The responsible-use note
Defeating AI detection does not change the underlying integrity question. If AI help is prohibited on a specific assignment, using a humanizer to pass detection is still a policy violation, just a harder one to catch. This page exists to help writers understand the tooling, not to encourage misconduct. Know your institution's rules, and disclose AI use where required.If AI assistance is permitted and you just want to verify your final text reads naturally, run it through Leap's free AI score checker first. It runs in your browser, needs no account, and highlights the sentences with the strongest AI writing signals. Treat the score as a signal, not proof, and never as the only basis for an accusation.