EQTY Lab, in collaboration with Intel and NVIDIA, has unveiled the Verifiable Compute AI framework, a groundbreaking hardware-based solution for governing and auditing AI workflows. This innovation aims to establish trust in AI systems by enhancing explainability, accountability, and security, thereby encouraging wider AI adoption and development.
Key Highlights of Verifiable Compute
Advanced Cryptographic Notary System
- Core Features:
- Patent-pending hardware-based cryptographic AI notary.
- Tamperproof record for every data object and computation in AI training and inference.
- Compliance and Policy Enforcement:
- Real-time adherence to AI business policies and sovereign regulations, including the EU AI Act.
- Rooted in next-gen hardware from Intel and NVIDIA for enhanced security.
Next-Generation Confidential AI Integration
- Intel’s Role:
- Integration with 5th Gen Intel® Xeon® Processors and Intel® Trust Domain Extensions (Intel® TDX).
- Enhanced trust zone with confidential VMs for secure AI operations.
- NVIDIA’s Contribution:
- Compatibility with NVIDIA H100/H200 GPUs and the forthcoming Blackwell GPU architecture.
- Setting a standard for cryptographically secure AI workloads.
Tackling AI Supply Chain Risks
- Addresses critical challenges like:
- AI poisoning and privacy backdoors.
- Information extraction and denial-of-service attacks.
- Ensures cryptographically secure records of every AI lifecycle stage, fortifying against supply chain vulnerabilities.
Compliance and Enterprise Integration
- Governance Gate:
- Halts non-compliant systems, notifying remediation tools.
- Integrates natively with platforms like ServiceNow, Databricks, and Palantir.
- Audit Capabilities:
- Issues AI audit and lineage certificates for instant verification or independent auditing.
- Long-term preservation of AI artifacts for regulatory conformity.
Market Potential
- Confidential computing demand is projected to grow exponentially, with market sales estimated at $184.5 billion by 2032.
- Compliance with emerging AI regulations and data sovereignty laws drives this demand.
Statements from Industry Leaders
- Jonathan Dotan, Founder, EQTY Lab:
“Verifiable Compute revolutionizes AI governance by combining cutting-edge cryptography with real-time compliance, enabling organizations to innovate responsibly.” - Anand Pashupathy, VP, Intel:
“This collaboration with EQTY Lab enhances the security, privacy, and accountability of AI solutions, driving trust in Confidential AI.” - Michael O’Connor, Chief Architect, NVIDIA:
“Securing the trust boundary in processors paves the way for cryptographically secure next-gen AI workloads.”
Verifiable Compute AI framework sets a new benchmark for AI safety and innovation by addressing escalating risks and compliance challenges. This collaboration between EQTY Lab, Intel, and NVIDIA empowers enterprises to innovate responsibly, bridging the trust gap in AI systems and preparing for the advent of autonomous AI agents.
