ShieldGemma is a series of security content auditing models built on Gemma 2, developed by Google, focusing on four harm categories: adult content, dangerous content, hate speech, and harassment. These are text-to-text decoder-only large language models available only in English, with open weights, including models with 2B, 9B, and 27B parameter sizes. These models are designed as part of a responsible generative AI toolkit to enhance the safety of AI applications.

Binoculars is an advanced AI-generated text detection tool that can be used zero-configuration without any training data. Its detection approach is straightforward and clear: most因果 language models that use only the decoder have used large amounts of identical datasets for pre-training, such as Common Crawl and Pile. More information about this method and its effectiveness can be found in our paper 'Finding LLMs with Binoculars: Zero-shot Detection of Machine-generated Text'.

Censorfy is a platform that utilizes artificial intelligence technology for instant content review, capable of reviewing text and images to ensure compliance. This service can be completed with a single HTTP request, providing users with an efficient and accurate content review solution. Key advantages of Censorfy include high efficiency, low cost, and user-friendliness, making it suitable for businesses and individuals that require rapid review of large volumes of content.

The omni-moderation-latest model is a next-generation multimodal content moderation system built on GPT-4o. It offers improved accuracy in detecting harmful information in both text and image content, aiding developers in building more robust moderation systems. This model supports both text and image inputs and demonstrates enhanced accuracy, particularly in non-English languages. It can assess whether content falls under categories such as hate speech, violence, and self-harm while providing more nuanced moderation decision controls. Additionally, it offers probability scores to reflect the likelihood of content matching detected categories. This model is freely available to all developers, aiming to help them benefit from the latest research and investments in safety systems.

The CNKI Personal AIGC Detection Service utilizes high-quality literature big data resources, structured, fragmented, and knowledge-optimized, combined with knowledge-enhanced AIGC detection technology and various detection algorithms. It detects AIGC-generated content from two dimensions: language patterns and semantic logic, using AI technology to help users quickly and accurately identify AI-generated content in academic texts. This service is significant for maintaining academic integrity and improving research quality.

The AI Content Detector is a tool that automatically identifies whether text content is written by a human or generated by AI. It utilizes advanced algorithms to analyze the grammar, word choice, and logical structure of the text to determine its origin. This tool is of great value in areas such as education, digital media, content creation, and marketing, enhancing the efficiency and accuracy of content review. The product supports API calls, allowing users to integrate it into their applications or systems easily.

HUMN-1 Certification is a website certification service launched by Winston AI, aimed at verifying whether website content is created by humans rather than generated by AI. This helps enhance user trust in website content, improve search engine rankings, and stand out in a web landscape filled with AI-generated content.

The CNKI personal plagiarism check service is a professional document checking service provided by CNKI. It leverages advanced technology to detect duplicate content within documents, enhancing the ability to identify automatic rewriting and machine-generated reduction. Additionally, the service has expanded its recognition of cited content, improving accuracy in identifying classical poetry, important speeches, regulatory texts, and case documents.

ShieldGemma is a series of security content auditing models built on Gemma 2, developed by Google, focusing on four harm categories: adult content, dangerous content, hate speech, and harassment. These are text-to-text decoder-only large language models available only in English, with open weights, including models with 2B, 9B, and 27B parameter sizes. These models are designed as part of a responsible generative AI toolkit to enhance the safety of AI applications.

Worry-Free Intelligent Review System is a text error checking and intelligent correction system based on big data, artificial intelligence (AI), and natural language processing technology (NLP). It can comprehensively proofread various types of text errors through deep learning, effectively improve the efficiency of manual review, eliminate blind spots in review, and enhance content security and text quality. The system supports various deployment methods, including embedded version, full-site review, and interface version, to meet the needs of different industries and scenarios.