SOVEREIGN DATA & AI COLLABORATION
Organizations need to collaborate on data to generate insights and build AI. But the most valuable data is also the most restricted.
Privacy, regulation, and security risks make sharing it difficult or impossible.
→ Data stays siloed. Value is never realized.
Organizations must collaborate across clouds, jurisdictions, and partners.
But most approaches force a tradeoff: collaborate and lose control, or retain control and operate in isolation.
→ Sovereignty must extend into collaboration
Duality enables organizations to compute on sensitive data and models without exposing them.
Instead of requiring raw data or models to be shared, approved computations are executed where they reside.
This enables collaboration across organizations, clouds, and environments while maintaining control.
Powered by privacy-enhancing technologies, including homomorphic encryption, trusted execution environments,
and federated learning.
• Data and models remain encrypted or protected
within their originating environments
• Policies define allowed computations and access
• Approved workloads execute across distributed
environments (cross-cloud, on-prem, hybrid)
• Only policy-approved results are revealed
• Customers retain full control of encryption keys, generated as ephemeral, PQ-secure keys within attested TEEs with no persistence in the cloud or reliance on the provider’s control plane.
→ Raw data and models are not exposed during
collaboration
Multi-Party Analysis
Enables joint analysis and benchmarking between organizations without exposing raw underlying data.
Distributed AI
Allows model training where the data resides and supports secure inference, protecting both the model and user data in use.
Cross-Boundary Operations
Supports compliant data use across different cloud providers, hybrid systems, and national jurisdictions without centralizing the data.
Compliant AI Controls
Provides technical controls, such as confidential computing, to satisfy regulatory and security requirements for AI data use.
Collaborate across agencies and partners on sensitive
intelligence data under strict policy controls
Analyze data across business units or partners
without centralizing or exposing sensitive information
