Scrydon

The National Security Agency has just issued a stark public warning: #MCP, the protocol powering most enterprise #AgenticAI is critically insecure. Its message is blunt. MCP’s explosive adoption has far outpaced its #security model. Authentication is optional, approval workflows cannot be enforced at protocol level, and essential controls such as RBAC, token lifecycle and audit logging are left entirely to implementers. Even MCP’s own documentation admits authorisation is “optional but strongly recommended” and requires complex external OAuth plumbing that still leaves the door open to the very exploits the NSA has now documented. For regulated sectors, such as #AIinDefence, #AIinFSI and #AIinGovernment this is no longer acceptable. At Scrydon we took a fundamentally different approach. Our Cognitive Enterprise Platform does not bolt security onto MCP. Every agentic workflow is defined upfront. The Human + AI Orchestrator then enforces identity, approvals, trust boundaries and just-in-time credentials by design, delivering deterministic, sovereign control with full immutable audit trails. https://lnkd.in/dHnsKiG3 If you’re deploying agentic AI in a high-stakes environment, drop me a message. Happy to discuss how Scrydon turns these NSA recommendations into production reality today.

The EU is about to write more rules on cloud sovereignty than it has working sovereign clouds. On 17 April the Commission awarded its first contracts under the new Cloud Sovereignty Framework. SEAL (Sovereignty Effectiveness Assurance Levels, 0 through 4) scores eight criteria from jurisdiction to supply chain. Four consortia will supply EU institutions with sovereign cloud services, up to €180M over six years. The SEAL grid finally turns sovereignty into something measurable instead of marketing. On May 27 the Tech Sovereignty Package lands. CNBC reports it will likely require member-state public bodies to process financial, judicial and health data on high-sovereignty infrastructure. The package pairs those demand-side rules with supply-side investment in cloud and chips. Three consortia reached SEAL-3 (Digital Resilience). Proximus reached only SEAL-2, leaning on S3NS, the Thales-Google Cloud joint venture. CISPE, the body of 38 European cloud firms, called it an own goal. A SEAL-2 service operated by EU staff on Google technology, under a contractual immunity regime, may deliver more reliable continuity for a Belgian hospital than a SEAL-3 provider with thirty engineers and a fragile next funding round. Sovereignty is a risk gradient, and the framework scores it that way. The framework cannot build the missing capacity. US hyperscalers hold around 70 percent of the European cloud market and European providers around 15. New rules on May 27 will not change those numbers. Procurement that consistently rewards SEAL-3 will. Multi-year contracts that survive ministerial cycles will. The risk is that May 27 produces more compliance language than capacity. The honest sequencing is to build the SEAL-3 supply, run the procurement signal long enough for it to compound, and only then write the rules that bite. #CloudSovereignty #EUTechPolicy #DigitalSovereignty #PublicSector sources: - EU Cloud Sovereignty Framework: https://lnkd.in/e6mtcUT3 - Commission award announcement (17 April 2026): https://lnkd.in/eK5yFDYT - CISPE response, reported by The Register (20 April 2026): https://lnkd.in/e6YmmTz3 - CNBC on the Tech Sovereignty Package (7 May 2026): https://lnkd.in/er332mmm

Just back from NATO TIDE Sprint 44 in Istanbul, and the message was unmistakable: AI and data aren’t future concepts. They’re the NOW of warfare. From Ukraine’s lessons to real-time sensor fusion and counter-drone orchestration, one thing became crystal clear: the Alliance that masters Human+AI collaboration at speed — while preserving sovereignty and interoperability — will hold the cognitive advantage. Key reflections from the week: AI-enabled data fusion is moving from theory to tactical reality Ontology-based standardization offers a real path through the “speed vs. interoperability” dilemma. Agentic AI with strong human guardrails and full auditability isn’t optional — it’s essential Sovereign platforms that deliver enterprise-grade data maturity first are the foundation for meaningful alliance sharing At Scrydon, we’re building exactly that: a sovereign Human+AI Orchestrator designed for defense — visual agentic workflows, air-gapped deployment, complete traceability, and multi-domain situational awareness. TIDE Sprint 44 didn’t just validate the direction. It confirmed the urgency. If you’re working on defense AI strategy, multi-domain operations, or sovereign capability development — I’d love to connect and exchange perspectives. Read the full reflections here: https://lnkd.in/eA_P6QxC #DefenceTech #NATO #AI #AgenticAI #SovereignTech NATO Allied Command Transformation (ACT)

🚨 Dans l’Aérospatial & Défense, une #IA non souveraine est une faille opérationnelle. #delaware est partenaire de Scrydon : une plateforme européenne d’IA agentique souveraine, conçue pour les environnements critiques et classifiés. ❌ Pas de données qui sortent ❌ Pas de dépendance cachée ❌ Pas de compromis sécurité / performance ✅ Agentic AI locale (RGPD & EU AI Act by design) ✅ Analytics haute performance, même en environnement déconnecté ✅ Data Spaces traçables et auditables Déployable airgapped, on-premise, cloud souverain - Azure uniquement si autorisé. Plus de contrôle. Plus de vitesse. Plus d’autonomie stratégique. 👉 La souveraineté n’est plus un débat. 👉 C’est une exigence opérationnelle. Tout savoir sur #Scrydon : https://lnkd.in/dRYFq6m6 #Aerospace #Defense #SovereignAI #CriticalSystems #AgenticAI #Scrydon #delaware #Souveraineté #IA #IntelligenceArtificielle #ERP #SAP #Microsoft #Aéronautique #Spatial #AD #Europe #UE