Developers introduced the text-humanizer tool to bypass AI detectors

Developers have released text-humanizer — an open-source tool designed to rework AI-generated text. The program uses a multilingual processing pipeline to alter the stylistics of the content in order to reduce the likelihood of its detection by detectors. The system's operation is based on integration with the DeepSeek model, which is responsible for rewriting text fragments. To ensure multilingualism and variability, the pipeline includes Google Translate and optional support for DeepL. The tool is distributed under a free license and is available for local deployment via a GitHub repository. The solution's architecture allows users to flexibly customize the processing process, combining various APIs to achieve the desired level of text "humanity". The use of multiple translators makes it possible to effectively change the sentence structure and lexical composition, which complicates the work of classification algorithms tuned to search for patterns of standard LLM. The tool is aimed at users who require automated adaptation of content for various platforms. The project has already gained 28 stars on GitHub within the first 8 hours after publication, which testifies to the high interest of the community in methods for bypassing AI text automatic detection systems.