Intel's Tiber AI: Confidential Federated Learning Redefined with Confidential Computing


Posted on: 18 Mar 2025 | Author: Foresiet
header

Introduction

As AI is reinventing industries, data security and privacy continue to be the most important challenges so far. Intel's Tiber Secure Federated AI service proposes addressing this by having a secure means of training AI without needing to move sensitive data.

Rather than filling AI models with data, Tiber reverses that—seeding fragments of AI models via a secure pipeline to communicate with data where it lives. That's especially critical in healthcare and finance, where data security needs to be paramount above all else.

How Tiber Secure Federated AI Works

Traditionally, AI models need huge amounts of data, which are being moved across networks, and there are threats of data exposure and compliance violations. Intel's Tiber service uses confidential computing, so data never leaves its source. This is how it works:

  1. Secure Tunnel Deployment: Tiber provides a secure communication tunnel between remote AI models and on-premises data sources.
  2. On-Site AI Training: Rather than shipping data, AI models train locally so that data never leaves its safe bubble.
  3. Model Updates Without Data Transfer: When training is done, the retrained model—no original data—is shipped back via a secure tunnel to the initial AI system.

This federated learning process allows organizations to work together on AI training without exchanging sensitive data.

Key Use Cases: A Game Changer for Regulated Industries

Intel's Tiber AI is well suited to companies requiring the best security and AI training offloaded to regulatory compliance requirements, i.e.:

  • Financial Services: Banks are able to build AI-trained models for detecting fraud without any access to customer information and thereby enhance security and collaboration.
  • Healthcare: Hospitals and organizations can build more advanced predictive analysis and diagnostics without access to patients' information.
  • Government and Defense: Organizations that handle sensitive information are now able to leverage AI innovation without placing data in restrictive control.

"Businesses can take advantage of next-gen AI breakthroughs while assisting in delivering data confidentiality and AI model trustworthiness of data for multi-party collaboration in distributed environments," says Rajan Panchanathan, Head of Product, Intel's Trust and Security Services.

The Role of Confidential Computing

Underpinned by Tiber's security architecture is Intel's Trusted Execution Environment (TDX), which provides an even more secure safe for processing data. Confidential computing does protect interactions of sensitive data with the model from access except by authorized systems and staff.

Security issues still exist, though, as prior breaches of confidential computing architecture foretell weaknesses that need to be continuously enhanced.

Challenges in Federated AI Training

While federated AI training offers significant advantages, it also comes with its own set of challenges:

  • Data Standardization and Quality: Standardization and quality of data within companies are the pillars of AI model performance.
  • Governance and Collaboration: Multi-party AI building requires precise agreements on objectives, security controls, and regulatory requirements.
  • Security Vulnerabilities: Previous security weaknesses in secure computing environments, such as attacks against TDX and AMD's SEV-SNP, highlight the ongoing need for improvement.

Binary CEO Alex Matrosov comments, "It's very challenging to truly have confidential computing when companies just layer on top of a broken-by-design stack. We need to go back and rebuild the building blocks first before we get to make that kind of flashy claim."

Conclusion

Intel's Tiber Secure Federated AI is an important milestone in securing AI training by using confidential computing to protect data while enabling federated AI co-development.

While there are obstacles, the fact that it provides a secure, federated learning platform makes it a viable option for industries handling sensitive data. As AI security continues to evolve, technologies like Tiber will play a key role in fostering a world where businesses can harness AI's potential without compromising data privacy.


About us!

Foresiet is the pioneering force in digital security solutions, offering the first integrated Digital Risk Protection SaaS platform. With 24x7x365 dark web monitoring and proactive threat intelligence, Foresiet safeguards against data breaches and intellectual property theft. Our robust suite includes brand protection, takedown services, and supply chain assessment, enhancing your organization's defense mechanisms. Attack surface management is a key component of our approach, ensuring comprehensive protection across all vulnerable points. Compliance is assured through adherence to ISO27001, NIST, GDPR, PCI, SOX, HIPAA, SAMA, CITC, and Third Party regulations. Additionally, our advanced antiphishing shield provides unparalleled protection against malicious emails. Trust Foresiet to empower your organization to navigate the digital landscape securely and confidently.

Safeguard Your Reputation, Data, and Systems

Protect your brand, reputation, data, and systems with Foresiet's Integrated Digital Risk Platform. 24/7/365 threat monitoring for total peace of mind.

dashboard