Map the AI estate
Discover approved and unregistered models, applications, agents, endpoints, datasets, knowledge sources, MCP servers and tools; connect ownership and dependencies in an AI bill of materials.
Sovereign AI security · Roadmap
Map, test and defend the AI estate against attack.
Discover AI assets and dependencies, verify models and tools before release, attack-test applications and agents, and contain hostile behaviour through security controls operated inside the customer-defined boundary.
Status: Roadmap. This application is being developed for open-source release. Its repository, OSI-approved software licence, model-specific terms, version, provenance and supported deployment modes will be published with the applicable release.
Core capabilities
Designed around accountable organisational work.
Exact coverage, deployment requirements and limitations are documented for the selected version and use case.
Verify the supply chain
Inspect repositories, model artefacts, containers, dependencies, MCP packages and tools for malicious code, unsafe deserialisation, hidden payloads, tampering, backdoors and vulnerable components.
Attack-test before release
Run repeatable adversarial tests against models, applications and agents for prompt injection, jailbreaks, extraction, tool misuse, privilege escalation, memory poisoning and resource abuse.
Defend live AI execution
Inspect model traffic, retrieved context, agent actions and tool calls for adversarial behaviour, then allow, block, rate-limit, quarantine, revoke or escalate through approved enforcement points.
Investigate, respond and evidence
Build event timelines across affected assets and dependencies; manage severity, remediation, exceptions, retesting and revocation; and export evidence to authorised security operations workflows.
Product boundary
Security, not ordinary application policy
AI Defense owns cybersecurity posture, hostile-component detection, adversarial testing, threat containment and investigation. AI Guardrails owns expected behaviour, structured outputs, quality evaluation and organisation-defined business policy.
Adoption outcome
Understand the attack surface before expanding AI
Begin with an asset and dependency inventory, define plausible attack scenarios, then establish security testing, release gates, containment and incident evidence appropriate to the environment.
Sovereignty matrix
Evaluate the deployment across six dimensions.
Product-specific answers are established through the deployment architecture and dated evidence record.
DataOperationalTechnologyJurisdictionKnowledgeExit
Review the trust model
Connected Roadmap capabilities
See how this responsibility fits the platform.
Every connected capability shown here is part of the current Roadmap.
SovereignEU AI Guardrails
Validate AI behaviour, evaluate quality and enforce customer-defined policies with local evidence.
SovereignEU Assistants
Permission-aware assistants, agents, workflows and connectors with human approvals.
SovereignEU Model Runtime
Customer-operated model serving, routing, versioning and resource management.
A practical first step
Understand the attack surface before expanding AI
Begin with an asset and dependency inventory, define plausible attack scenarios, then establish security testing, release gates, containment and incident evidence appropriate to the environment.