AI sovereignty, provenance and compliance
summary
Defines AI sovereignty as reclaiming control and giving concrete steps: cryptographically signed models, dataset provenance receipts and SBOM/AI BOM inventories, plus reproducible builds and lockfiles. Demonstrates tool patterns (model cards, Sigstore-style signing, Delta Lake / Iceberg lineage, Prompt Fu or DeepEval for repeatable evaluation), runtime choices from airgapped to gatewayed deployments, and observability for detection. Claims these interventions are implementable on a laptop and stresses automated CI/CD checks so artifacts can be verified and reproduced.
AI-written summary · about
chapters (24)
- 0:00Introduction
- 1:48AI sovereignty and threat model
- 3:13Castle and moat metaphor
- 5:41Autonomy in AI systems
- 6:44Key sovereignty principles
- 7:25Provenance in AI systems
- 8:29Threat model across AI stack
- 9:11Data layer threats
- 10:35Environment threats and supply chain
- 11:39Agentic systems attacks
- 13:03Data layer provenance interventions
- 15:15Model provenance and verification
- 16:42Cloud sovereignty framework
- 18:08Artifact registration
- 19:30Runtime choices as decisions
- 21:37Environment interventions and SBOMs
- 22:41Reproducible builds
- 23:27Evaluation layer testing
- 24:51Evaluation cards and experiments
- 25:58Integration layer and AI gateways
- 27:23Sustainable autonomy
- 28:49Self-hosted reality check
- 30:36Supply chain security
- 32:47Q&A
related talks
24mBuilding Sovereignty: Where Your AI Runs Matters
33mAI in the Security Stack
23mTrust, Track, and Verify: Securing AI Pipelines End-to-End
28mthe AI Software Supply Chain
20mWrap-Up Session Panel Discussion: Sovereign AI
44mAI, Corporate Responsibility & Democratic Legitimacy
19mWe Know Security but How Do We Secure GenAI End-to-End?
08mWhy AI Governance Matters More Than Ever