"The key principle is simple: Legal AI must serve the law."

Europe's legal systems are among the most sophisticated in the world – but they face growing pressure from rising caseloads, complexity, and resource constraints. At the same time, the question of how AI can support legal work is becoming increasingly urgent, especially when professional secrecy, institutional independence, and democratic accountability are at stake.

Against this backdrop, Noxtua is building a Legal AI platform that deliberately prioritizes sovereignty: domain-specific models, curated legal data, and infrastructure operated entirely under European law. Dr. Leif-Nissen Lundbæk, CEO and Co-Founder, is driving this approach from Berlin – a city where legal tradition, tech innovation, and regulatory expertise converge.

In this conversation, he explains why Legal AI must be built differently from generic tools, how partnerships with leading European legal publishers enable jurisdiction-specific intelligence, and what role sovereign infrastructure plays in protecting professional secrecy and the rule of law. He also discusses the specific requirements of courts and legal professionals, the boundaries that responsible Legal AI must respect, and how European collaboration across borders strengthens AI development for the legal domain.

Leif, your platform is designed to support and streamline day-to-day legal work; how does Noxtua assist legal professionals in practice today, and what are the key principles behind the AI system it is built on? 

Noxtua is designed as a Legal AI Workspace. In practice, that means supporting legal professionals across the full workflow, from research to analysis to drafting and editing legal documents. Legal professionals use Noxtua to find relevant regulations more quickly, structure complex issues, analyze large document sets, and create legally sound drafts – all with clear references to authoritative sources. 

The key principle behind the system is simple: Legal AI must serve the law. That means we do not replace legal professionals, but we make them more efficient. That requires a domain-specific model architecture trained on curated legal data, full transparency of sources, and an infrastructure that guarantees confidentiality and professional secrecy. We deliberately avoid generic, internet-trained models and instead build AI that understands legal language and reasoning. 

We develop the Legal AI Workspace in close partnerships with leading European legal publishers across the continent, making each version specifically tailored to the respective countries and jurisdictions. In Germany, we work closely with C.H.BECK, Germany's largest legal publisher, for Beck-Noxtua. 

Noxtua is clearly positioned as a Legal AI for professional use; which user groups are you primarily addressing at the moment, and how do their needs and use cases differ? 

Our primary user groups today are law firms, corporate legal departments, public authorities, and the judiciary in Europe. While their use cases differ, their shared core requirements are accuracy, trust, confidentiality, and legal compliance. 

For the judiciary in particular, these requirements are not optional – they are constitutional. Courts are one of the most sensitive institutions of a democratic state. Any AI system used in this context must therefore be fully controllable under European law, transparent in its reasoning, and operated on sovereign European infrastructure. 

This is why sovereign Legal AI matters so much for the justice system. Courts are under growing pressure from rising caseloads and structural staff shortages. Properly designed Legal AI can support them immensely by structuring information, handling routine tasks, and increasing efficiency, while legal professionals remain fully in control. 

A core promise of Noxtua is faster and more precise legal research based on reliable sources; how do you ensure the quality and relevance of your legal content, and how transparent are AI-generated results for users in practice? 

Legal AI is only as good as the data it's based on. That's why we built Noxtua on curated, jurisdiction-specific legal content from leading European publishers, like C.H.BECK in Germany, MANZ in Austria, and other trusted partners across Europe. Each country version is based on the local legal system and curated legal data, enabling Noxtua to reflect the professional standards of each jurisdiction. 

Equally critical is that Noxtua is developed together with legal professionals. The Legal AI Workspace aligns with real workflows, not abstract technical assumptions. It works the way legal professionals think and work. 

And most importantly, users can verify which commentary, statute, or court decision an answer is based on. Without that, Legal AI has no place in serious legal practice. 

Legal work involves highly sensitive and confidential information; how does Noxtua approach data protection, professional secrecy, and security in everyday use, and what role does European infrastructure play in this context? 

We built Noxtua on sovereign European infrastructure. We do not rely on US hyperscalers, and user data is neither permanently stored nor used for model training. In Germany, Beck-Noxtua runs on highly secure European cloud infrastructure provided by IONOS and the Open Telekom Cloud (OTC). Together with IONOS, we even initiated Germany's first sovereign Legal AI Factory in Munich. All data processing takes place on certified European servers, without any connection to non-European cloud providers. 

From a technical perspective, we combine end-to-end encryption with confidential computing on trusted execution environments. Organizationally, our security and governance are anchored in internationally recognized standards, including ISO 27001, BSI C5, TISAX, and ISO 42001 for responsible AI governance. 

European infrastructure matters because professional secrecy only exists if it is enforceable. Where foreign jurisdictions can access data, confidentiality becomes a promise without substance. That is why Beck-Noxtua fully complies with the strict requirements of professional secrecy under German law (e.g. Section 203 German Criminal Code, Section 43e German Federal Code for Lawyers), and other Noxtua country versions comply with comparable standards across Europe. 

Noxtua was founded in Berlin and today operates across several European locations. How did this development come about, and what role does the European context play in shaping Noxtua’s vision and positioning? 

The initial idea sparked in Oxford and London – we founded the company in 2017 in Berlin, and have since opened offices in Paris, Zagreb, and Munich. That's Europe in a nutshell! Noxtua is a genuine European AI project, aiming to promote Europe's sovereignty – especially in these times of increasing geopolitical uncertainty. The legal domain specifically needs to ensure its independence and sovereignty, as the rule of law is one important pillar of our European democracies. 

With the launch of Beck-Noxtua, you entered a close partnership with Germany’s leading legal publisher C.H. Beck. How does this collaboration change the way legal knowledge is accessed and used in AI systems, and what have you learned from combining a traditional legal publisher with an AI startup? 

The partnership with C.H.BECK is fundamental to Noxtua. C.H.BECK isn't just a content provider; they're also our lead investor and strategic partner. What makes this partnership transformative is that we embed AI directly into a content-based AI Workspace. Beck-Noxtua is built on beck-online data – more than 60 million curated legal documents covering all areas of German law, with a strong focus on commentary literature that's essential for legal reasoning. No other AI provider has access to this content, now or in the future. This changes how Noxtua works: it learns legal reasoning from authoritative sources and references them transparently. 

What we've learned from this collaboration is that tradition and innovation aren't opposites. They're mutually reinforcing when both sides share the same standards for quality and responsibility. That's the foundation Beck-Noxtua is built on. 

As Legal AI tools become more widely adopted, what do you see as the most important prerequisites for their responsible use in legal practice, and where do platforms like Noxtua need to draw clear boundaries? 

There are two main prerequisites: human accountability and sovereign control. Legal professionals must remain responsible for legal judgment. And the AI itself must be fully operated and controlled under European law. 

That requires clear boundaries. Legal AI must not rely on generic models, foreign infrastructure, or uncontrolled data flows. And it must never weaken professional secrecy or institutional independence. An AI system can only be called sovereign if models, data, and infrastructure remain under European control. Where this is not the case, Legal AI becomes a structural risk rather than a tool for the rule of law.

Thank you for the conversation.