Trust & Privacy
Comprehensive security, privacy, and compliance for enterprise AI applications
Overview
Welcome to the Trust Center for VecML. Our commitment spans security, privacy, resiliency, and compliance, ensuring your data is protected, available, and handled responsibly.
Our platform is designed to meet the rigorous demands of modern AI and data-centric applications. Whether you're building real-time recommendation engines, semantic search, or large-scale analytics, we provide the foundation you can depend on.
Committed to Safeguarding Customer Data
At VecML, protecting customer data is a top priority. We continually enhance our security practices and controls while maintaining transparency in how data is processed. Our commitment includes meeting the highest standards of compliance to support even the most demanding security and privacy requirements.
Trust Is Ongoing
We know trust is earned every day. Our Trust Center is your window into our practices and our promise to uphold them. If you have specific questions, need documentation for due diligence, or report any potential security and privacy issues, please contact us directly at:
Security
Our security is designed to protect data in every phase: at rest, in transit, and during query execution. We follow industry-leading standards and adopt a multi-layered defense strategy.
Key Security Features
Encryption
AES-256 encryption for data at rest; TLS 1.2+ for data in transit
Authentication & Access Control
Support for RBAC, fine-grained permissions, API keys, and OAuth 2.0
Audit Logging
Tamper-resistant logs for API activity, authentication events, and configuration changes
Secure Development Lifecycle
Regular scanning, threat modeling, pen testing, and manual review
Querying with AI Database of Encrypted Attributes and Vectors
Encrypted Attribute Filtering
We support filtering on encrypted metadata (e.g., age, category, geolocation) enabling precise queries while maintaining data confidentiality.
Encrypted Vector Similarity Search
We allow similarity search over encrypted vector embeddings through distance-preserving encryption, maintaining search accuracy while protecting sensitive data.
Use Cases
Healthcare
Retrieve similar patient records while preserving PHI
Finance
Conduct anti-fraud vector searches without leaking transaction metadata
Retail & Ads
Recommend products based on user behavior embeddings, securely
Defense & Intelligence
Run entity or similarity analysis over sensitive surveillance data
Privacy
We take a privacy-first approach to handling personal and customer data. Our architecture and policies are aligned with global data protection laws.
Privacy Principles
Data Minimization
We only collect the data necessary to deliver and improve our services.
User Control
Customers retain full control over their vector data, metadata, and backups.
Transparency
We clearly communicate how we handle your data in our Privacy Policy.
Data Residency
Options for regional data storage and processing
GDPR & CCPA Ready
Support for data subject rights (access, delete, portability)
Unmatched DP for Vector Embedding
We provide industry-leading support for applying Differential Privacy (DP) to vector embeddings—enabling secure AI applications that minimize information leakage while preserving semantic utility.
The DP technique used is based on our research paper:https://arxiv.org/pdf/2306.01751
Video demo for training neural nets to achieve much higher testing accuracy (and much faster) than standard neural net package. 0.9723 vs 0.8995 to retain excellent learning accuracy under mathematically rigorous privacy protection at ε =2, 3, 4, 5.
Performance Breakthrough
Previously DP (w/ ε≤10) only has a bad accuracy of 65%. Our novel contribution has improved the accuracy to 92%.
This breakthrough enables practical deployment of differentially private machine learning while maintaining the utility needed for real-world applications.
Resiliency
VecML resiliency strategy centers around a tightly integrated Security Information and Event Management (SIEM) system that continuously monitors infrastructure, data pipelines, and application behavior. Our SIEM solution is deeply embedded across the stack—from storage and query engines to API gateways and orchestration layers.
Key SIEM Capabilities
Real-Time Anomaly Detection
Tracks behavioral baselines for query patterns, indexing throughput, node health, and latency. Deviations trigger automated investigations.
Threat-Informed Monitoring
Uses known failure models to contextualize signals from vector database nodes and disk I/Os.
Automated Response
When incidents are identified, pre-defined playbooks trigger mitigation workflows, such as automatic failover, rate limiting, or isolation of affected components.
Unified View
Operators and engineering teams access a centralized dashboard for correlated alerts, audit logs, and incident timelines, facilitating swift root cause analysis.
Compliance
Through independent audits and certifications, VecML ensures that its security and privacy controls align with the stringent standards required by global regulations and industry-specific policies.
Compliance Standards
SOC 2 Type II
Annual audits of security, availability, and confidentiality controls
ISO 27001
Information security management system certification
HIPAA-Readiness
Controls to support PHI-handling use cases
Data Protection Agreements
Standardized DPA templates
Third-Party Risk Management
Continuous vetting of cloud and subservice providers
Compliance Ready
Questions About Our Trust & Privacy Practices?
Contact our trust team for detailed documentation, compliance certifications, or security inquiries.