Got Questions? We've Got Answers.

Our AI detection model achieves 99.2% accuracy on benchmark datasets. It uses a multi-model ensemble approach trained on content from ChatGPT, Claude, Gemini, and other major AI systems. False positive rates are kept extremely low through calibration.

We can detect content generated by ChatGPT (GPT-3.5, GPT-4, GPT-4o), Claude (Anthropic), Gemini (Google), Llama, Mistral, and most other widely-used large language models. We continuously update our models as new AI systems are released.

We recommend at least 150 words for reliable detection. Very short texts (fewer than 50 words) may produce less reliable results as there is not enough context for the model to make a confident prediction.

Minor edits (changing a few words, adding punctuation) typically do not defeat our detection. Our model analyzes statistical patterns across the entire document, not just individual words. However, heavy paraphrasing or our own Humanizer tool will significantly reduce the AI probability score.

Your text is broken into sentence-level segments and compared against our database of over 100 billion web pages, academic publications, and publicly available documents. Matching passages are highlighted and linked to their original sources.

No. Your submitted text is never stored in our plagiarism index and will never appear as a source in future checks. We use your text only to perform the analysis and generate the report. See our Privacy Policy for full details.

We cover publicly accessible web content including many academic publications. Access to paywalled databases (such as JSTOR, Elsevier, or Springer) is limited by their access policies. We continue to expand our coverage partnerships.

Our Humanizer significantly reduces AI probability scores across major detectors. However, no tool guarantees 100% evasion since AI detectors are constantly evolving. We recommend reviewing the humanized output for quality and using it as a starting point for your own editing.

Yes, our model is optimized to maintain the core meaning, facts, and structure of your original text while rewriting the style and phrasing to sound more natural. We provide a side-by-side view so you can compare the original and humanized versions.

Yes. You can upgrade your plan immediately from your account settings — new quota takes effect right away. Downgrades take effect at the end of your current billing cycle.

Yes, we offer custom pricing for schools, universities, and organizations. Contact us at contact@textauthentic.com with your institution name and expected number of users and we'll prepare a tailored quote.

Yes. All data is transmitted over HTTPS and encrypted at rest. We do not sell or share your personal data with third parties. Our infrastructure is hosted in the EU, compliant with GDPR requirements.

Paid plan users can access their full analysis history for as long as their account is active. You can delete individual analyses or your entire history at any time from your dashboard. On the Free plan, history is not retained.