ScribeLens is an AI-writing detector at scribelensai.com. Paste or upload a document and it returns an overall verdict, a sentence-by-sentence breakdown, and an exportable report — built around showing the evidence behind a result, not just a number.
You paste text or upload a PDF, DOCX, TXT, or Markdown file, and the whole document is analysed in one pass. Every sentence is classified individually as human-like, mixed, or AI-like and highlighted in the document, underneath a graded overall verdict: likely human-written, mixed signals, a blend of human and AI writing, substantial AI involvement, or strong signs of AI writing. The headline is always built to agree with the percentages printed beside it.
You can run a quick check with no account, on up to 2,000 words per scan. A free account gives you 35,000 words a month and 9 PDF/DOCX report exports, and it scans a whole essay or paper in one go.
A bare percentage doesn't tell you what to do next. Sentences that get flagged carry reason keywords naming the specific pattern behind the flag — even sentence rhythm, a generic transition, formulaic structure — so a result is something you can actually inspect and act on, not just react to.
AI writing leaves measurable statistical patterns, and ScribeLens is built to detect and surface exactly those patterns, sentence by sentence. What a result can never be is proof: detection is probabilistic evidence for a person to weigh, not a verdict to hide behind in either direction.
Disciplined academic prose can read as AI-like for two concrete reasons, and ScribeLens handles both. Reference lists are the most formulaic text in any paper by construction, and detectors that score them read that rigidity as machine-like — ScribeLens excludes reference lists, in-text citation apparatus, DOI/ISSN lines, copyright and licence text, and declaration sections (Author Contributions, Funding, Ethics statements, and similar back-matter) automatically before scoring. Formal register — consistent structure, measured transitions, even sentence rhythm — is also exactly what detectors are trained to flag, so ScribeLens weighs sentence-level evidence against document-level style signals rather than convicting a paper on register alone.
We built both fixes by testing against published, peer-reviewed journal articles during development and tracing every case where a genuine paper was misclassified back to its cause — that's what drove the exclusion rules and the sentence-weighting above, not a guess at what academic writing 'should' look like.
Excluded content is still shown in the report for transparency; it just never affects the verdict.
A single accuracy number can't describe how a detector performs across every document type, subject, length and writing style. So instead of a headline figure, ScribeLens shows you the evidence behind each result: the flagged sentences, the pattern named for each one, and the detector version that produced them.
We name patterns, not products: a result shows where AI-like patterns appear, not which tool wrote the text, and we retrain the detector as writing tools change. We focus on one job, AI-writing detection, and do it in depth.
The live version is Detector v4.0. Researchers can evaluate it independently through our research access, and we welcome published findings, including ones that don't flatter us.
Students checking a draft before they submit it, educators reviewing student work, researchers and academics scanning full papers, and journal editors screening submissions with evidence they can share — that's who ScribeLens is built for.
It also suits people who work with AI text professionally: LLM trainers and evaluators reviewing model output, prompt engineers and AI content reviewers checking how strongly a draft carries AI-style patterns, and technical writers auditing text before it ships. Sentence-level evidence with named reasons is exactly the granularity that kind of review needs — and the API, with keys provisioned by agreement, supports it at volume for teams.
ScribeLens is built and operated by the ScribeLens team. For questions, feedback, or institutional access, reach us at support@scribelensai.com.
ScribeLens is built and operated by the ScribeLens team, reachable at support@scribelensai.com.
Sentence-level evidence: every sentence classified and highlighted, reason keywords naming the exact pattern behind each flag, and an exportable report anyone can read. You judge the result with the evidence in front of you.
No — we don't reduce detection to a single unverifiable percentage. Every verdict shows you which sentences were flagged, and the reason keyword behind each flag, so you can check the result against the text yourself. A quick check is free and needs no account.
No. ScribeLens detects the patterns that AI assistants share, so a result shows where AI-like patterns appear in a text, not which tool produced it. We retrain the detector as writing tools change.
Email support@scribelensai.com.
Paste or upload your writing and read the sentence-level evidence behind the result. A quick check of up to 2,000 words needs no account; a free account scans the whole document.