ReText.AI AI Detector Update: Now Detects AI-Generated Text with Higher Accuracy
What’s New
In this version, we’ve improved the algorithm that analyzes the probability of AI generation.
That means the detector now identifies AI-generated sentences and paragraphs with greater precision — especially in long texts (over 2,000 characters) and in English-language content.
The AI Detector now also better recognizes the writing styles of different models and GPT versions — it can distinguish ChatGPT from Claude, Gemini, and GigaChat.
We further trained the neural network to detect paraphrased texts more accurately — previously one of the most challenging tasks for any detector.
We achieved this major leap in accuracy by updating the transformer model and expanding the training dataset with new examples of AI generations from 2025.
The detector now evaluates context on a deeper level — not just individual sentences, but also the logical structure and flow of the text.
Testing Results
We conducted research that was later validated through the AINL Eval competition — one of the largest benchmarks for AI detectors designed to identify machine-generated content.
ReText.AI ranked among the leading teams alongside Sber.
At the international PAN 2025 evaluation, our detector was also recognized as one of the most accurate.
Based on the test results:
- The average accuracy in detecting AI-generated texts increased by 12%, and by 25% for English texts;
- False positives on human-written content decreased by 18%;
- Long-text analysis became four times faster.
Who Benefits Most
Our AI Detector helps you find out who actually wrote the text — a human or a neural network.
For example:
- Teachers use it to check student papers;
- Copywriters rely on it to ensure the originality of commissioned articles;
- SEO specialists monitor website content to prevent AI-generated material from affecting search rankings;
- Companies of all sizes use the detector to verify the authenticity of documents and maintain trust.
What’s Next
We continue to enhance our service and train our models so you can create better texts anytime you need. Soon, we’ll introduce several more important improvements. If you have suggestions for improving our AI Detector, we’d love to hear from you — message us on Telegram or email us at team@retext.ai.
Our AI Detector evolves alongside modern neural networks. It accurately identifies AI-generated content, helping you maintain transparency and trust in every text. Try the updated AI Detector for free at ReText.AI.