pds-it
['Blog post','no']
Microsoft Technology
Blog

Microsoft expands Azure with Azure AI Content Safety

Contents

    Azure AI Content Safety is a new AI service from Microsoft. It is designed to create safe online spaces by recognizing inappropriate content. customers can set themselves what is inappropriate and how Azure AI Content Safety should deal with such content. The application is based on modern AI models from OpenAI.

    What Azure AI Content Safety can do

    Microsoft's Azure AI Content Safety is already available. The service is based on a modern AI from OpenAI, which is able to recognize certain content in images and texts. Specifically, this involves inappropriate and offensive content. For example, offensive, violent, sexual or hurtful content.

    There are setting options within Azure AI Content Safety. This allows user:in to set the severity level for certain content. This makes it possible to create a limit for content and prioritize content for moderation review. user:in can therefore set themselves what goes through the filter and what does not. This is particularly useful for online spaces in companies.

    The number of false positives from Azure AI Content Safety should be low as the application captures contexts and nuances.

    The individual functions of Azure AI Content Safety are currently as follows:

    - Classification of unsafe content: Azure AI Content Safety divides inappropriate content into the four categories "violent", "self-harming", "hate" and "sexual".

    - Severity levels for content: The scale ranges from 0 to 7. The higher the severity level, the more inappropriate the content logically is.

    - Semantic understanding of texts: Using natural language processing techniques, the AI recognizes both the meaning and context of a text. It should therefore imitate human intelligence.

    - Real-time recognition: Instant content recognition means that nothing slips through by mistake.

    - Multilingual: Azure AI Content Safety understands English, German, Japanese, Spanish, French, Italian, Portuguese and Chinese.

    It is to be expected that the service will be expanded and improved over time. How well context recognition works, for example, remains to be seen. This is an area that AI has always had problems with and certainly has room for improvement.

    Another new function in addition to Azure AI Content Safety

    In addition to Azure AI Content Safety, Microsoft is expanding the scope of Azure to include a feature for model monitoring for generative AI applications. The feature is part of Azure Machine Learning.

    With model monitoring, companies record changes to the data and the behavior of customers . In the long term, these changes ensure that AI applications operate with outdated data sets, which inevitably leads to poorer results and high costs.

    Model monitoring for generative AI applications solves this problem. Continuous monitoring of data and performance indicators ensures that the system is always up to date.

    Using AI in Azure with skill it

    Azure AI Content Safety and model monitoring for generative AI applications are just two of the many features, functions and applications within Microsoft Azure. The cloud is huge and offers countless opportunities to work with artificial intelligence.

    In the Designing and Implementing a Microsoft Azure AI Solution training , you will learn about some of the functions of Azure in connection with AI in twelve modules and get hands-on experience in practical exercises. The course requires knowledge of the programming languages C# or Python and is aimed at software developers who are interested in creating applications with Microsoft Azure.

    Author
    Marcel Michaelsen
    Marcel writes IT content for websites as a freelancer at Textflamme. The topics range from product descriptions to complex technical articles.