Paste your draft
Sign in and add a passage of at least 50 words. Check your remaining word allowance before a new scan, and use only writing you are permitted to submit.
Review AI, mixed, and human probabilities alongside sentence-level signals. Save a check in your account and improve your draft with the humanizer.
Free and paid accounts · 50-word minimum · Uses your shared word allowance
Highly likely AI-generated
These example probabilities illustrate the report layout. They are not measured results, an accuracy benchmark, or a percentage of AI-written words.
Illustrative sentence highlights
Sentence scores are separate estimates and can differ from the overall document classification.
In today’s rapidly evolving landscape, innovative solutions are essential. A first draft can help a team get started. I still read the final email aloud before sending it.
50+
words to analyze
3
probability categories
Account-only
result archive
How it works
An AI detector can help you review a piece of writing. Start with the text, read the result in context, and decide what deserves another edit.
Sign in and add a passage of at least 50 words. Check your remaining word allowance before a new scan, and use only writing you are permitted to submit.
Review the verdict and all three category probabilities, then inspect the sentence highlights. A low fully-AI probability can still appear with a high mixed probability; read both before drawing a conclusion.
Improve clarity, specificity, and voice with your own edits or the WriteBros humanizer. Review the rewritten text before another check.
Inside the report
The detector brings the overall assessment, sentence signals, and saved checks into the same workspace.
The report distinguishes fully AI-generated, mixed AI and human, and human-written probabilities. Mixed or uncertain signals need context. These are model classifications, not proof of who wrote the text.
Highlighted passages help you inspect where the detector finds stronger or weaker AI signals. Use them to review your wording, rather than treating each highlighted sentence as a verified authorship label.
Your recent checks appear inside your account with the verdict, date, word count, and text preview. Open a saved result when you need to revisit a draft or compare your own revisions.
The detector and humanizer share the same WriteBros workspace and word balance. A changed draft may need a fresh scan, so leave enough words for both rewriting and checking.
Read the draft’s signals
Edit clarity and writing voice
Review the revised result
Writing in context
Review a blog, newsletter, or social draft before publishing. Keep your facts, examples, and personal perspective at the center; a detector score does not establish originality or content quality.
Inspect a permitted client draft or business document alongside your style guide. Check confidentiality before submitting or sharing text, and judge the final writing against the brief.
Use detection as a prompt to review your own writing process, citations, and clarity. Follow your institution’s AI policy. A result cannot prove academic integrity or guarantee how another system will classify your work.
FAQ
Signed-in free and paid accounts can use the detector when their account is eligible and enough words remain. New scans use the same word allowance as humanization. A past-due paid account needs its payment issue resolved before a new scan.
Usage is measured in input words, not a fixed number of monthly checks. A fresh scan spends its input word count from your shared allowance. Free allowance is lifetime; paid allowance follows the billing period. When an cached result matches the trimmed input text, it is returned without another word charge. Edits may require a new scan.
The detector accepts passages of at least 50 words. The passage must also fit the words available in your shared account balance for a new scan. Short excerpts do not satisfy the minimum.
No. The fully-AI, mixed, and human figures are category probabilities, not percentages of words written by AI. Detectors can misclassify text. Use the result alongside writing history, sources, and human judgment; it is not a definitive authorship or misconduct finding.
No. This report classifies writing signals; it does not verify or certify removal of a specific watermark such as OpenAI textGrain. An AI-likelihood result and a watermark detection result are different tests.
Create a free account or sign in, check your remaining words, and open the AI Detector. Compare paid plans if your writing workflow needs a larger shared word allowance.