Paperpal AI detector

The free box reads "Start typing or paste your text here (supports up to 1,200 words for free)", and its upload tooltip accepts a ".docx up to 25 MB" (paperpal.com/tools/ai-detector, read 10 November 2026). The comparison table states the free tier as "5 scans/day, 1,200 words each". The FAQ states the paid allowances: "Prime gives you 25,000 words of AI detection per month, and Pro gives you unlimited usage", matching the pricing page's feature lines of "25k words / mo AI Detection" and "Unlimited AI Detection"; the Free column there reads "Preview AI detection". No per-scan ceiling is stated for paid plans and no API is documented.

The pricing page renders its prices by script from a subscription packages API. The figures below are the USD display amounts that API returns for a United States visitor on 10 November 2026 (paperpal.com/pricing and its packages response); the response's base price field differs for two packs (139 against the $144 displayed for annual Prime, 63 against the $59 displayed for monthly Pro), and the card records what is displayed.

The packages data also lists 10,000 words a month of AI detection on the Free package, a figure no page displays; the card leaves the free monthly word field empty and ranks the free tier on its stated 5 scans a day. Retention is stated on three pages. The detector page says: "Your sensitive academic work stays secure with Paperpal. We don't store your drafts or use them to train our AI models." The Data Security FAQ reads: "Documents you create: delete them whenever you want. Files uploaded for plagiarism and other checks: delete whenever you want. Files uploaded for Journal Fit checks: automatically deleted after 90 days. Personal data: retained while your account is active and as required by legal obligations" (paperpal.com/data-security, read 10 November 2026). The privacy policy keeps personal information "as long as an account is active or as needed to provide the service(s)" and states no period for submitted text. The card's retention field takes the detector page's no-storage line as a stated zero and quotes the other two.

The Data Security page names standards: ISO/IEC 27001:2022 and ISO/IEC 42001:2023 certifications, GDPR, FERPA, PECR and a CSA STAR Level 1 listing, with data in "SSAE 18/SOC1-certified facilities" (paperpal.com/data-security, read 10 November 2026). Paperpal runs as an MS Word add-in, a Chrome and Edge extension for Google Docs and Overleaf, and a web app.

The FAQ states: "Paperpal AI detector can identify AI-generated text from Gemini, Claude, and ChatGPT with 99% (accuracy), providing a nuanced AI checker report" (paperpal.com/tools/ai-detector, read 10 November 2026). The page body adds that the tool is "Built on 100,000+ human and AI scholarly samples from ChatGPT, Claude, and Gemini" and "outperforms generic AI checkers on detection accuracy", and that "Paperpal's multi-model checks reduce false positives by over 40%, so original writing is not judged by a single AI detector score". No dataset, count or date is given for the 99%, and the 40% is a relative reduction with no base, so no false positive rate is recorded.

The same FAQ carries the caveats. "While today's AI checkers are increasingly sophisticated, no AI content detector can guarantee 100% accuracy. Most assign a probability or confidence score that indicates the likelihood of AI-generated text. Accuracy may vary depending on the model used, writing style, and whether the text has been edited or paraphrased by a human." On paraphrase: "Yes, it's possible for paraphrased text to be flagged by AI detectors", and on evidence: "Academic institutions typically treat AI detection reports as supporting evidence, not definitive proof of misconduct" (paperpal.com/tools/ai-detector, read 10 November 2026). The vendor sells an AI Paraphraser on the same site and recommends it in the FAQ for rewording drafts, so the humanized text field quotes both the detection claim and the paraphraser.