Frequently asked questions
Accuracy, languages, privacy and access - the things people ask before they trust a detector.
Accuracy & detection
Are AI detectors accurate?
No detector is perfect. Review the methodology, supported languages and known limitations, and treat every score as evidence to investigate rather than standalone proof.
Why do false positives happen?
Detectors read statistical patterns, and some human writing - formulaic, templated or non-native - can look machine-like. Sentence-level evidence helps people review flags fairly and in context.
Which AI models can you detect?
Detection is model-agnostic: GPT-4o, Gemini, Claude, Copilot, Llama, Mistral and more - including paraphrased and humanized text.
How is this different from plagiarism detection?
Plagiarism tools compare text to existing sources. AI detection estimates whether text was machine-generated. They answer different questions and are complementary.
Languages
Which languages do you support?
English, Spanish, French, Italian, Portuguese and Romanian - each detected natively rather than translated into English first.
Why does native detection matter?
Translating text before analysis can erase linguistic signals detection relies on. Native evaluation preserves more of the language-specific context.
Privacy & security
Is my text kept private?
Yes. Detection runs in-house on Google Cloud. Your text is never shared with third parties and is not used to train models.
Are you GDPR and Law 25 compliant?
We align with GDPR and Quebec’s Law 25, support data-subject rights, and offer a DPA and data-residency options for enterprise customers.
Access & API
How do I get access?
Contact our team with your use case, expected volume and languages. We will recommend the right next step.
Do you offer an API?
Yes - detection is available via API for publishers, platforms and agencies. See the API page to request access.
Detect AI with confidence - in any language
Tell us about your use case, volume, language, and privacy requirements.