About
WasItAIGenerated detects AI-generated text, images, video and audio. Teachers, editors, fact-checkers, investigators and platform teams use it on our website, through our API and inside AI agents.
Every result is a probability from a model. We say so on every result and never present a score as proof about a person.
We publish our benchmarks with the data behind them, keep a hidden test split to stop overfitting, and invite outside researchers to audit us.
Uploaded files are deleted after analysis and never used to train our models. Enterprise customers can run the models on their own servers.

Founder
Patrick has worked on natural language processing since his bachelor's thesis in 2021. He holds an M.Sc. in Data Engineering and Consulting from Hochschule Albstadt-Sigmaringen, where his thesis built an embedding-based anomaly detection system for a production SaaS application.
Before WasItAIGenerated he built several software products on transformer models and vector search. He started WasItAIGenerated to give people a detector they can check, with published benchmarks and an open-source text detector.
2026
A new detector for images and video clips, open benchmarks on GitHub and Hugging Face, and tropa-3 for text.
2024
WasItAIGenerated goes live with text, image and audio detection on the website and through an API.
2023
First detection models, built and tested against the generators people were using at the time.
2022
The project starts from a simple question: how do you tell whether a text or image was made by AI?
Check a text, image, video or voice recording for free, or ask about the API and enterprise pilots.