Connect with us

Hi, what are you looking for?

Top Stories

Microsoft Expands Foundry with Gated Hugging Face Models for Secure AI Deployment

Microsoft enhances Azure with gated Hugging Face models, enabling secure AI deployment across regulated sectors like education and government.

Microsoft has expanded its AI model catalog in Microsoft Foundry to incorporate gated open-source models from Hugging Face. This move enables organizations to deploy advanced models within Azure while maintaining tighter controls over access and usage. The update is a response to a broader trend in enterprise AI adoption, as companies increasingly prioritize governance, accountability, and responsible deployment, particularly in regulated sectors such as education, government, and platforms focused on children.

Jessica Hawk, Corporate Vice President of Azure Product Marketing at Microsoft, emphasized the importance of safety-driven use cases in a recent LinkedIn post. Hawk’s comments reflect growing concerns about the need for controlled access to AI models, highlighting the shift from open access to governed deployment.

According to Microsoft, gated models are placed behind explicit access boundaries defined by model publishers, requiring users to request and gain approval before deployment. This process is seamlessly integrated with Hugging Face user access tokens, allowing organizations to adhere to licensing and usage conditions without the burden of manual approval workflows. This approach aims to provide enterprises with a wider range of high-quality open-source models while ensuring necessary oversight.

In her post, Hawk cited Roblox as a case study illustrating why governed access is essential at scale. Roblox processes billions of chat messages and utilizes a specialized PII Classifier model to flag personal information in real-time. Notably, this model is now available through Microsoft Foundry, showcasing the practical implications of these gated models.

The rollout of gated models will occur in phases, beginning with a diverse array of tools focused on computer vision, language, and safety. Initial offerings include the Segment Anything Model 3 (SAM 3) from Meta, which supports prompt-based object segmentation applicable in fields such as medical imaging and content moderation. The Roblox PII Classifier, designed to detect personally identifiable information (PII) within chat messages, and FLUX.1 Schnell from Black Forest Labs, a model optimized for rapid, high-quality image generation, are also part of the initial release. Other models include EuroLLM-9B-Instruct, which supports over 30 European languages, and Bielik-11B-v3.0-Instruct, an instruction-tuned model excelling in multilingual reasoning, particularly in Polish.

Microsoft has stated that organizations can deploy these models directly into secure online endpoints within Foundry once access is validated. This capability is particularly significant for education technology providers and systems, as there is an increasing demand for AI tools that are both powerful and governable. Such tools are vital in environments involving students, minors, or sensitive data. As AI skill development continues to evolve, the emphasis on model selection, deployment, and oversight positions platforms like Foundry as integral to responsible AI usage rather than merely facilitators.

The ongoing evolution of AI governance and model access reflects broader industry trends towards responsible AI practices, emphasizing the balance between innovation and accountability. As enterprises navigate the complexities of deploying robust AI solutions, Microsoft’s initiatives may serve as a benchmark for others in the field.

The ETIH Innovation Awards 2026 are currently open, recognizing education technology organizations that demonstrate measurable impact across K–12, higher education, and lifelong learning sectors. The awards welcome entries from the UK, the Americas, and internationally, with submissions evaluated based on the evidence of outcomes and real-world applications.

See also
Staff
Written By

The AiPressa Staff team brings you comprehensive coverage of the artificial intelligence industry, including breaking news, research developments, business trends, and policy updates. Our mission is to keep you informed about the rapidly evolving world of AI technology.

You May Also Like

AI Business

Red Hat advances enterprise AI with Small Language Models that achieve over 98% validity in structured tasks, prioritizing reliability and data sovereignty.

AI Cybersecurity

Anthropic's Mythos exposes thousands of critical vulnerabilities in major systems, prompting $100M in defensive action from tech giants and U.S. banks.

AI Government

US Department of Defense partners with tech giants including SpaceX and OpenAI to launch an "AI-first" initiative aimed at enhancing military decision-making efficiency.

AI Research

OpenAI's o1 model achieves 81.6% diagnostic accuracy in emergency situations, surpassing human doctors and signaling a major shift in medical practice.

AI Regulation

Korea Venture Investment Corp. unveils AI-driven fund management systems by integrating Nvidia H200 GPUs to enhance efficiency and support unicorn growth.

AI Technology

Apple raises Mac mini starting price to $799 amid AI-driven inventory shortages, eliminating the $599 model in response to surging demand for advanced computing.

AI Research

IBM launches a Chicago Quantum Hub to create 750 AI jobs and expands its MIT partnership to advance quantum computing and AI integration.

AI Government

71% of Australian employees use generative AI daily, but only 36% trust its implementation, highlighting urgent calls for better policy frameworks and safeguards.

© 2025 AIPressa · Part of Buzzora Media · All rights reserved. This website provides general news and educational content for informational purposes only. While we strive for accuracy, we do not guarantee the completeness or reliability of the information presented. The content should not be considered professional advice of any kind. Readers are encouraged to verify facts and consult appropriate experts when needed. We are not responsible for any loss or inconvenience resulting from the use of information on this site. Some images used on this website are generated with artificial intelligence and are illustrative in nature. They may not accurately represent the products, people, or events described in the articles.