AI Disclosure

This disclosure covers AI use in GPTHuman product features and editorial work. It explains the role of automation; it does not mean that every page, decision, or sentence was generated by AI.

Editors may use AI to explore outlines, identify unclear phrasing, summarize notes they already possess, prepare a translation draft, or suggest alternative wording. These outputs are working material, not publishable evidence.

AI-generated claims, quotations, examples, links, and citations must be checked against reliable sources. An editor may rewrite, reject, or replace any output that is inaccurate, generic, unsupported, or inconsistent with the intended meaning.

An accountable editor reviews the final structure, material facts, links, examples, claims, and disclosures before publication. AI is not credited as a human author and is never the sole source for a factual statement.

The named contributor or GPTHuman Editorial Team remains responsible for what readers see, including AI-assisted passages. Publication approval, corrections, and judgment calls belong to people.

Our editorial workflow prohibits:

• Unreviewed AI output, fabricated quotations, or citations that have not been opened and verified.
• Invented people, credentials, customer stories, test results, statistics, or sources.
• Mass-produced pages that repeat the same information mainly to attract search traffic.
• Claims that mislead readers about authorship, detector accuracy, guaranteed rankings, or guaranteed outcomes.

The current Humanizer, AI Detector, AI Writer, and Grammar Improver send submitted content and instructions to Groq's API so a configured model can return a result. The Readability Scorer performs rule-based calculations in the application and does not send the text to that AI provider.

GPTHuman's application database records operational usage data such as the tool, word count, timestamp, and the account or guest identifier needed to enforce quotas; it is not designed to store the submitted text or generated result in usage history. The AI provider and hosting infrastructure may process request data under their own configured logging and retention terms. Do not submit confidential or regulated information unless you are authorized to do so.

AI output can be incomplete, biased, outdated, or wrong. Detector scores are estimates, not proof of authorship, and rewriting cannot guarantee a particular detector score, search ranking, academic decision, or platform outcome. Users must review facts, citations, originality, and the rules that apply to their work.

We add a specific disclosure when AI use materially affects an article's evidence, experiment, translation status, or likely reader interpretation. Routine spelling, grammar, formatting, and outline assistance do not normally receive a separate label, but they still require human review.

If we learn that AI-assisted material introduced a material error or unsupported claim, we review the source, correct or remove the passage, and follow the correction process in our Editorial Policy.