Gerwin AI is now almost secure. We have added a new filter that is designed to detect generated text that may be confidential or insecure. It is currently in beta mode.
Currently it is more difficult to analyze complexly formatted text, so if it has many line breaks, unusual formatting, repeating words and so on, the model might misclassify it. At this point, the filter also reduces performance for certain types of text, such as fiction, poetry, etc.