Unlock Collaborative AI with Hardware-Level Privacy
The open-source foundation for confidential computing. Train and run AI models on sensitive data within secure enclaves, ensuring complete data sovereignty and trust.
Trusted Partners
Collaborating with leading organizations in confidential computing and secure AI.
Memberships
Proud members of key industry consortia.
How It Works
Secure Multi-Party Computation made simple and scalable.
1. Secure Enclaves
Data and code are isolated in hardware-encrypted Trusted Execution Environments (TEEs) like AMD SEV.
2. Remote Attestation
Cryptographic proof ensures that the workload is running on genuine hardware and hasn't been tampered with.
3. Confidential Computing
Compute on encrypted data without ever exposing it to the cloud provider or infrastructure owner.
Confidential Computing Consortium
Cocos AI is an innovative project funded by the Serbian government’s Innovation Fund, developed by:
Ultraviolet
Specialized high-tech services in cybersecurity, focusing on confidential computing in cloud, telecom, and AI/ML.
University of Belgrade
School of Electrical Engineering, a leading academic institution in the region.
Use Cases
Real-world applications of Cocos AI's confidential computing capabilities.
Medical / Pharma
- Medical trials-encrypted patient data exchange for identifying new treatments and drugs
- Drug distribution, serialization, and tracking
Public / Government
- Analytics on citizen data, or AI/ML on top of confidential data coming from different agencies
Banking / Finance
- Private client payment data exchange for payment fraud detection and anti-money-laundering
- Credit approval process handling in a secure manner
- Transaction handling between two parties
Transportation / Logistics
- Support for autonomous driving (routing, AI/ML…)
- Logistics Data Analytics
Industry
- Multi-party computation on data from competing and/or cooperating companies
- AI/ML for predictive maintenance
- Quality Assurance and Data exchange
- Production Data Exchange (Industry 4.0)
Software Development
- Intellectual property protection (algorithms protected from cloud provider)
- Insider threats (employees within a company)
Key Features
Powerful capabilities for secure and efficient AI workloads.
Secure VM Provisioning
Seamlessly deploy and maintain secure Virtual Machines.
Hardware Abstraction Layer
Unified interface for AMD SEV and Intel TDX.
In-Enclave Agent
Execution scheduler and coordinator within secure enclaves.
End-to-End Encryption
Encrypted traffic safeguarding confidentiality and integrity.
Remote Attestation
Verify the integrity of remote systems and establish trust.
Multiple Runtimes
Support for Python, Docker, and WebAssembly workloads.
SDK and CLI
Comprehensive tools to streamline development.
High Performance
Small memory footprint and lightning-fast execution.
Open Source
Apache 2.0 license, fostering transparency.
About Us
Cocos AI is an innovative project funded by the Serbian government’s Innovation Fund.
Ultraviolet
A company with a reputation in the security domain, coordinating the project and building the user experience and SaaS offering. Ultraviolet ensures a high-quality product for the market.
Visit WebsiteETF University of Belgrade
The top academic institution in the country. Experts in data security and protection bring novel research and IP into the product, focusing on TEEs and secure multi-party computation protocols.
Visit WebsiteThe project is funded by the Innovation Fund from the budget of the Republic of Serbia from the division of the Ministry of Education, Science and Technological Development (Project ID 50314).
Contact Us
Reach out to us for more information or collaboration opportunities.
ETF University of Belgrade
Bulevar Kralja Aleksandra 73
11120 Belgrade
Serbia