Microsoft launches tools to try and stop people messing with chatbots

Machine Learning AI
(Image credit: Shutterstock)

Microsoft has launched a suite of new tools within Azure aimed at boosting the safety and security of generative AI applications, with a special focus on chatbots.

The tools are hoped to help organizations mitigate various risks associated with the deployment of generative AI, including concerns about abusive content and prompt injections.

The latest offerings include features like real-time monitoring to track and shut down abusive content or users as well as protections against new attack vectors like jailbreaks and prompt injections.

Microsoft boosts chatbot security

According to a recent McKinsey survey, nine in 10 (91%) corporate leaders express feeling unprepared for the risks associated with generative AI despite the benefits it could have on their business in terms of productivity and efficiency. 

Microsoft’s new tools aim to address these concerns, and are the culmination of technical innovation and research based on its own experience with in-house products like Copilot.

Microsoft’s multibillion-dollar investment in OpenAI has undoubtedly played an important role in this, unlocking various opportunities for research and experience in the AI landscape.

Prompt injections involve manipulating AI systems to generate harmful unintended content. Microsoft’s Prompt Shields is designed to block both direct and indirect prompt attacks. The tool relies on advanced machine learning algorithms and natural language processing to analyze prompts and third-party data for potentially malicious intent.

As well as addressing security concerns, Microsoft’s latest tools are also hoped to improve the reliability of generative AI applications by automatically evaluating them through stress testing to ensure that risks like jailbreaks are minimized.

Another prominent update, real-time monitoring, is designed to let developers track inputs and outputs that trigger safety features, allowing them to tweak the back end and adjust content filter configurations to improve safety more manually.

It seems that every AI-related announcement Microsoft has made in recent months has affirmed its commitment to responsible and safe AI, and these latest tools are certainly a testament to that.

More from TechRadar Pro

Craig Hale

With several years’ experience freelancing in tech and automotive circles, Craig’s specific interests lie in technology that is designed to better our lives, including AI and ML, productivity aids, and smart fitness. He is also passionate about cars and the decarbonisation of personal transportation. As an avid bargain-hunter, you can be sure that any deal Craig finds is top value!