Thursday, May 25, 2023
As part of the Microsoft Build Azure AI Content Safety introduced a new AI service on Microsoft Azure to help you create safe online spaces. Using state-of-the-art AI models, this service can detect and assign severity levels to abusive, violent, sexual, and self-harm content in images and text, allowing organizations to limit and prioritize what content needs to be reviewed by moderators.
Unlike most solutions in use today, Azure AI Content Safety can capture nuance and context, reducing the number of false positives and reducing the burden on human moderation teams.
Azure Content Safety features
- Classification of unsafe content: Azure AI Content Safety classifies malicious content into four categories: sexual, violent, self-harm, hate.
- degrees of severity: The service provides a severity level on a scale of 1 to 6 for each unsafe content category.
- semantic understanding: AI-driven content moderation uses natural language processing techniques to capture the meaning and context of speech and mimic human intelligence. It can analyze both short and long texts.
- Multilingual models: The service understands several languages: English, German, Japanese, Spanish, French, Italian, Portuguese and Chinese.
- Customizable settings and regulatory compliance: Customizable settings to comply with regulations and policies.
- Computer Vision: The service is based on Microsoft’s Florence basic model for performing advanced image recognition. This technology has been trained on billions of text-image pairs.
- real-time detection: The platform detects malicious content in real time.
Azure Content Safety represents a significant step in online security and provides organizations with an effective way to protect their platforms from inappropriate content.
Additional resources on Azure AI Content Safety
The most important announcements of the Microsoft Build
We have collected the announcements of this year’s Microsoft Build on a Special page on TechWiese put together for you. On the Microsoft Build Germany on June 22nd in Munich our experts illuminate the most important announcements in a series of lectures.