Alibaba Cloud AI Guardrails is a pioneer in AI application protection and content moderation. Leveraging Alibaba’s decades of technological expertise and deeply integrating the capabilities of the Tongyi large model family, it has established an AI security governance framework and a UGC content moderation system tailored to the AI era. Based on Aigc and UGC Content Management and Risk Protection Experience Accumulated in E-Commerce, Social Networking, Education, Games, Basic Models, AI Applications and Other Scenarios, It Provides One-Stop Risk Identification and AI Protection Services Covering Multi-Modal Content Such as Text, Pictures, Videos, Audio, and Documents. Provide Stable, Compliant, and Ready-to-Use AI Security Solutions for Enterprises and Developers to Efficiently Prevent Risks Such as Violations, False, Attacks, and Harmful Information, Comprehensively Improve the Security and Content Quality of AI Applications, and Help Build a Clear and Reliable Network Space.
Benefits
-
Leading Algorithm Capability
Based on Leading Algorithm Capabilities Such as Alibaba Group, Damo Academy and Tongyi Laboratory, Automatic and Accurate Identification Is Performed and Results Are Returned in Milliseconds.
-
Comprehensive Service
Supports Multi-Modal Video, Image, Text, and Audio, Covering Content Review, Prompt Attack, Sensitive Data, Model Illusion, and AI Identification.
-
Proven Performance
It Serves Customers in Social Networking, Live Broadcast, E-Commerce, Education, Aigc and Other Industries, and Has Rich Experience in Sample Accumulation and Control.
-
Customizable
AI Security Expert Team, One-to-One Algorithm Operation, Supports Personalized Effect Adjustment, and Completes Fast Iteration through Data Backflow.
Features
-
Guardrails
Risk Detection Capability
Covers Risk Scenarios Such as Content Compliance, Sensitive Data, Prompt Attacks, Malicious Files, Malicious URLs, Model Illusion, and Prompt Crawlers, and Supports Digital Watermark Embedding of Generated Content.
Custom Protection Configuration
You Can Change Refined Risk Detection Items in the Protection Configuration, Including Custom Detection Items, Custom Risk Thresholds, and Custom Filter Words.
Access Method
Supports API, AI Gateway, WAF, Bai Lian Model, Bai Lian Agent, Dify Agent, Openclaw Plug-in and Other Access Methods.
-
Content Security 2.0
Image Moderation 2.0
Support the detection of red line content in images, such as pornography, sexiness, violence, prohibition, flags, inappropriate, abusive, and special elements, including the content of the images and the text content in the images (supporting 18 languages such as Chinese, English, French, Russian, Japanese, Arabic, etc.).
Video Moderation 2.0
Detects Whether There Are Illegal Or Inappropriate Content in Video Files Or Live Video Streams. You Can Detect Images in Videos and Voice in Videos. We Recommend That You Perform This Detection on All Video Content That Involve Open Internet Access.
Text Moderation 2.0
Adopting Independent Strategies and Labeling Systems Can Effectively Identify Text Content Such as Pornography, Violence, Contraband, Advertising Drainage, Desecration and Abuse, and Regional Opposition. Supports 38 Language Types Including Chinese, English, French, Thai, Japanese, Korean, Russian, Portuguese, and Alibor. Provides More Features to Simplify Business Use and Assist Manual Review.
Voice Moderation 2.0
By Upgrading the Core Engine of Content Security, Voice Moderation 2.0 Provides Audit Services for Business Scenarios Such as Graphic Sharing, Game Linking, and Live Courses, Identifies Content Or Elements That Violate Network Content Dissemination Regulations, Affect Platform Order, and User Experience, and Provides Rich Content Risk Tags.
Document Moderation 2.0
Check Whether the Document Contains Image Or Text Violation Information, Including Pornographic, Sexy, Political, Terrorist, Prohibited and Other Bottom Line Content.
-
Content Moderation 1.0
Image Moderation
Through the Neural Network Algorithm and the Real-Time Updated 100 Million-Level Image Sample Library, You Can Identify the Risk Content of the Image, and Support the Detection of Yellow, Sensitive Information, Violent Violations, Bad Content, Logo and Other Dimensions.
Image OCR Service
Based on the Industry-Leading Deep Learning Technology, after Years of Polishing Various Business Scenarios, It Can Provide Users with Text Recognition Services for Various Scenarios. OCR Supports Recognition of Simplified Chinese, Common Traditional Chinese, and English.
Face Detection and Retrieval
You Can Detect Faces in Images, Compare Faces, and Search Faces.
Vedio Moderation
Through the Neural Network Algorithm and the Real-Time Updated 100 Million-Level Image Sample Library, the Video Is Identified to Detect Whether There Are Risk Content Such as Pornography, Sensitivity, Violence and Prohibited, Bad, Logo, Etc.
Text Anti-Spam Detection Service
Detects undesirable scenes in a video.
Upgraded Support For You
1 on 1 Presale Consultation, 24/7 Technical Support, Faster Response, and More Free Tickets.
1 on 1 Presale Consultation
24/7 Technical Support
6 Free Tickets per Quarter
