Zhuque AI Detection

Tencent has launched a free AI-generated content detection tool. Trained on a variety of advanced models and millions of data points, it delivers particularly outstanding performance in Chinese detection. It supports both text and image detection and provides an API for easy integration into review processes.

AI Detector#Detection#Tencent#Content Safety

Updated: 2026-10-26

What is Zhuque AI Detection?

Zhuque AI Detection Assistant is a free AI-generated content detection tool launched by Tencent. According to official introductions, it is mainly used to help users identify AI-generated text and image content. Trained on millions of pieces of data based on a variety of advanced artificial intelligence models, it can distinguish writing patterns between humans and AI. The official particularly emphasized that it not only has excellent English detection capabilities but also performs particularly well when processing Chinese data—it is especially valuable for self-media, academic, and review scenarios dominated by Chinese content. At present, its API capabilities have also been launched, allowing integration into self-owned processes. As a public detection capability launched by Tencent, it is targeted at various scenarios requiring quick judgment of content sources. The official lists text analysis and image/video analysis as core capabilities, and free online use minimizes the threshold for access.

Who is it for?

Teams that conduct pre-publication self-checks for self-media and official account authors, AI content verification for teachers and academic scenarios, AI content identification and data quality audits for content platforms, as well as evaluators who want to assess the capabilities of various detection tools; students can also use it as a reference when checking for AI writing traces (specifically subject to the school's regulations).

Key features

Text Detection

Identifies the probability that a text was generated by AI, covers the output styles of common large models, and the official examples simultaneously display AI-generated and human-written texts for comparison.

Image/Video Detection

Supports image detection in addition to text. Images can be uploaded directly for detection. The video detection capability is in the official's plan, making it suitable for multimedia content review.

Paragraph-by-Paragraph Analysis

Highlights suspicious paragraphs to facilitate manual review, rather than just providing a general conclusion, enabling more direct problem localization.

Scenario Coverage

The official lists typical scenarios such as social media AI content recognition, academic integrity, and data quality audit assistants, covering both content platforms and educational institutions.

Benchmark Evaluation

Provides public evaluations such as "Battle of the Strongest AI Identifiers", enabling horizontal comparison of the detection capabilities of various tools to assist in selection.

API Capability

The official API has been launched, making it convenient to integrate the detection capability into your own review and quality control processes.

Free to Use

Basic online detection is free, with low barriers to entry and ready-to-use access.

How to use Zhuque AI Detection

Detection is based on the judgment of statistical features, which may lead to misjudgments for human-written content (especially highly literary text).
The results should only serve as a reference rather than the sole basis.
Before batch access, it is recommended to first run a small sample with your own real content, observe the false positive rate, and then determine the threshold and process.
The official sample texts can help you quickly understand the judgment tendency.
it is recommended to prioritize Chinese content for detection, and English content can be cross-verified with other tools to improve confidence.

Things to know

Please look for Tencent's official domain. There are many third-party websites with the same name on the Internet that charge fees under the guise of the official name, so please verify the entry carefully. The test results cannot replace manual judgment. For serious scenarios involving academic integrity, content review, etc., it is recommended to make a comprehensive judgment by combining other evidence to avoid drawing conclusions based on a single tool; before formal access, also pay attention to the terms of service and data description. The detection model and threshold may be adjusted with version updates, and the test results of the same text may vary at different times.