Question 1

What is AI Text Detector?

Accepted Answer

AI Text Detector is a public text-checking form that returns a document-level estimate and explanation about a submitted passage. It is intended to support reading and review, not to certify a person’s identity or writing process. The service sends the processed text to an external model provider. Its availability does not establish a measured classification accuracy, a language ranking or an independently validated sentence-level result.

Question 2

How does AI Text Detector work?

Accepted Answer

The website accepts text, applies the implementation’s request controls and forwards the processed passage to an external analysis provider. It displays the returned estimate and explanation. This operational description is different from a published model architecture or benchmark. We do not claim a proprietary weighted ensemble, a particular perplexity formula or confidence intervals that have not been established by the inspected implementation. Read the how-it-works and accuracy pages for the boundary of the evidence.

Question 3

Is AI Text Detector free?

Accepted Answer

The public checker is available without an account or payment, subject to request controls. Its displayed price remains free; the earlier descriptions of paid Pro or Enterprise services should not be treated as active products. No unlimited-use promise is made. If the form declines a request or a provider is unavailable, that is an operational state, not an assessment that the submitted passage is human-written or AI-generated.

Question 4

Who is AI Text Detector built for?

Accepted Answer

Writers, educators, students and editors can use a permitted passage check as one part of reviewing a draft. A free public form is not automatically approved for a school’s student records, an employer’s confidential material or unpublished client work. The person submitting the passage needs permission for that sharing. Use the relevant workflow to determine whether a detector is appropriate before choosing which passage to submit.

Question 5

How accurate is AI Text Detector?

Accepted Answer

No verified accuracy rate or false-positive percentage is published here. The tool’s operation has been checked, but a successful response is not an accuracy experiment. A defensible rate requires retained material with known authorship, a defined task, recorded conditions and complete outputs, including failures. Do not treat an older numerical claim or a precise-looking output as a substitute for that evaluation.

Question 6

Can it detect text from ChatGPT, Claude, and Gemini?

Accepted Answer

The service returns an estimate about submitted text, but this site has not published a version-by-version performance evaluation for those model families. The name of a writing model does not establish that its output will be correctly classified. A comparison would need the actual version, prompt, passage and editing history, along with the detector conditions. It should not infer performance on a new model from a result on an older one.

Question 7

What about edited AI content?

Accepted Answer

Changing the wording creates a different input. A later score therefore need not match the earlier output, and the direction of a change does not reveal the complete writing process. Preserve the original passage and report when the distinction matters. Review whether revisions improve the argument, evidence and attribution rather than repeatedly editing to obtain a preferred label from a model.

Question 8

What about false positives?

Accepted Answer

A false positive occurs when human-written text is labelled as AI-generated. Published research documents such errors, including concerns about non-native English writing in the tested conditions. This does not provide a current error percentage for this site. If a result is disputed, read the passage, verify the sources and consider relevant process evidence through the applicable review procedure. Do not impose a consequence solely because of the label.

Question 9

Is my text stored after analysis?

Accepted Answer

The analysis involves an external provider. This site has not established an instant-deletion guarantee or a complete provider-retention audit, so do not submit sensitive material on the assumption that it cannot be retained. Read the current privacy and processing notices before sharing a passage. Preserve only the material you are authorised to keep, and use the approved institutional route if the document is subject to a school or workplace policy.

Question 10

Do you use submitted text to train your models?

Accepted Answer

This page does not promise a provider-wide training exclusion that has not been verified. The public form relies on external processing, and a reliable answer depends on the relevant provider terms and configuration. If training use, confidentiality or a retention period is a requirement, obtain a documented answer through the appropriate process before submitting the text. The absence of an account requirement does not answer those questions.

Question 11

Is AI Text Detector certified for my organisation’s privacy requirements?

Accepted Answer

No certification or organisation-specific approval is established by this FAQ. A working public form and a privacy page do not constitute a security review, data-processing agreement or approval to share confidential records. Use the organisation’s procurement and privacy process when those requirements apply. The contact page can help identify an unresolved service question, but do not infer a compliance guarantee from general product wording.

Question 12

What is the minimum and maximum text length?

Accepted Answer

The input field accepts up to 50,000 characters, while the current backend forwards only the first 15,000 characters for analysis. These are character counts, not word counts. This FAQ does not establish a universal minimum for reliable detection. Follow the form’s actual validation and keep a record of the processed excerpt. A result for the opening of a long document should not be presented as a validated finding about the remainder.

Question 13

What languages are supported?

Accepted Answer

Language-specific accuracy has not been independently established here. Do not rely on the old language-count or relative-accuracy claims as evidence for a particular document. A response in a language does not prove that classification performs well in that language. If language coverage is essential, obtain appropriate documentation and an evaluation that resembles the writing you intend to review before using results in a consequential process.

Question 14

Can I analyse multiple documents at once?

Accepted Answer

The established public workflow is a passage pasted into the form. A batch-upload product or paid batch service is not established by this review. Do not assume that a file count or a document-upload feature described in an old article is available. For several documents, identify the approved process and the actual input boundaries. Avoid losing track of which passage produced which output.

Question 15

Do you offer API access?

Accepted Answer

A public customer developer API is not established for this site. An internal request used by the website is not the same as a supported API product with authentication, documentation, service terms and support. The earlier paid-plan and API-rate claims should not be used to design an integration. If a developer workflow is required, obtain current product documentation from the relevant provider before implementation.

Question 16

What AI models can it detect?

Accepted Answer

No list of writing-model versions with measured detection performance is established here. Treat an output as an estimate for the supplied passage, not confirmation that a named model was used. Version changes, prompts and editing are conditions to record in a genuine evaluation. This site does not promise a retraining deadline after each model release or claim that a general-purpose model is equally accurate across every writing model.

Question 17

How is AI detection different from plagiarism detection?

Accepted Answer

AI detection estimates whether text has patterns associated with generated writing. Source matching looks for related wording in other material. A passage can contain a source-use problem regardless of its AI estimate, and an AI label does not identify a copied source. This site provides a separate source-overlap checker. Inspect the actual matches, quotation and attribution rather than treating either kind of output as a complete authorship judgment.

Question 18

Should AI detection results be used as sole evidence of AI use?

Accepted Answer

No. Use the output alongside the original text, sources, permitted process evidence and the writer’s explanation. A school or workplace should follow its own review procedure and provide an opportunity to respond. Do not replace the original submission with edited text and then use the later scan as though it described the original. Keep observations and inferences separate in any review note.

Question 19

How should I interpret the scores?

Accepted Answer

Read the returned explanation and identify the exact passage that was processed. A displayed score does not establish the proportion of sentences written by a model, the probability that a particular person used AI, or the chance of misconduct. This FAQ does not assign universal low, medium or high verdict bands. Ask what the report field means and which conclusion the available evidence actually supports.

Question 20

Does editing AI-generated text affect detection?

Accepted Answer

An edited passage is a new input, so a changed output is possible. This FAQ does not promise that editing always lowers a score or proves a particular composition history. Review revisions for meaning, factual accuracy and attribution. If the aim is to document the work, preserve the earlier draft and its output instead of replacing them with a sequence of scans selected for a preferred result.

Frequently Asked Questions

General