Microsoft Unveils Trustworthy AI
Microsoft has unveiled a suite of new AI security features, aimed at addressing growing concerns about AI security, privacy, and reliability.
The tech giant is calling the initiative “Trustworthy AI,” signaling a push toward responsible development of AI technologies.
Microsoft's announcement comes at a time when businesses and organizations are embracing AI solutions, bringing both opportunities and challenges.
New offerings from Microsoft include Azure OpenAI secret inference, improved GPU security, and improved tools for evaluating AI output.
Microsoft has unveiled Correction, a service that aims to address the problem of AI hallucinations, which are instances where AI models generate false or misleading information.
Correction attempts to review AI-generated text that is actually wrong, flags text that may be wrong, and then verifies its accuracy by comparing the text to sources.
Correction, which is available as part of Microsoft's Azure AI Content Safety API, can be used with any text-generating AI model, including Meta's Llama and OpenAI's GPT-4o.
Correction relies on a novel process of using small language models and large language models to align the output with underlying documents.
“When we detect a mismatch between the underlying knowledge context and the response, we feed that information back to the AI system,” the company said. “With this additional information, the AI system is typically able to provide better responses on the second try.”
Microsoft is also expanding its efforts in embedded content safety, allowing AI safety checks to run directly across devices, even when they are offline.
This feature is linked to applications that integrate AI capabilities directly into the operating system, such as Copilot.
The company's push toward trustworthy AI reflects a growing awareness in the industry of the potential risks associated with advanced AI systems.
This also puts Microsoft in a leading position in the development of responsible AI, giving it a competitive advantage in the cloud computing and AI services market.
The company's approach appears to be resonating with some high-profile clients, highlighting collaborations with school districts in New York City and South Australia.
These departments use Azure AI Content Safety to create relevant AI-powered learning tools.