Artificial Intelligence
BlackVogel works on the governance layer of AI.
Not model development. The questions around it. Who is accountable when a system acts, which rules apply, and whether frameworks written for human actors still hold when the actor is not human.
Our work runs across four areas:
research, standard setting, EU policy processes, and advocacy.
Research:
Accountability in agentic AI
AI agents are beginning to act and transact autonomously, including delegating tasks to other agents. Once that happens, the central question shifts from capability to accountability. Who is responsible when an autonomous agent causes harm.
We work at the legal and governance layer of that question, across three strands.
01
Liability in autonomous agent-to-agent delegation
When one agent delegates to another and the result causes harm, the chain of responsibility does not match the frameworks we have. Contract, product liability and tort were built for a different structure.
We map where responsibility actually lands, and where the loss cannot flow back to the party that caused it.
The core insight is that the problem is not the absence of liability. It is misallocation. Liability settles on the wrong party, and it cannot reach the actual source of the harm.
02
Best-practice design for terms governing autonomous delegation
We translate that analysis into design principles and clause language for the terms under which agents select and engage sub-agents.
Drafted to be enforceable under EU law, and usable by autonomous systems.
03
Comparative regulatory analysis
How existing EU frameworks handle analogous accountability problems, and what transfers to agent-commerce markets.
A template for the standards that will need to follow.
HAFTBAR:
Haftung, Adjudikation und Formalisierung für Transparente Beweisführung in Agentischen Rechtsbeziehungen. A funded research collaboration.
This research is not theoretical positioning. It is being carried out inside a funded German R&D project run with galaniprojects, a Berlin-based research and technology company.
The project addresses accountability in autonomous multi-agent markets: how agents that transact and delegate without human intervention can be held to account, and what mechanisms, collateral-based sanctions, dispute resolution between agents, enforceable terms of engagement, make that possible in practice.
BlackVogel leads the legal and regulatory research strand. Our contribution runs across the three areas above, and feeds directly into the technical design work rather than sitting alongside it. The engineering questions and the liability questions are answered together, which is the only way either answer holds.
The project is structured as collaborative research under the German Forschungszulagengesetz, the federal R&D tax incentive framework.
AI and blockchain convergence
Mariana de la Roche Wills co-chairs the AI & Blockchain Convergences Task Force at INATBA, the International Association for Trusted Blockchain Applications.
The task force examines what happens where the two technologies meet, from data provenance and transparency to security and ethical deployment, and turns that into strategic insight and actionable recommendations for the ecosystem and for regulators.
More information
Work delivered through the task force includes:
- INATBA’s report on AI and blockchain convergences
- AI Regulation and Blockchain: Bridging Ethics and Governance
- Positioning Progress: The Ecosystem’s Insights on AI and Blockchain Convergence
- INATBA’s reply to the European Commission call for contributions on competition in virtual worlds and generative AI
- An open, live repository of convergence use cases, open to contributions from any project
- A public webinar series
EU policy processes
We contributed to the drawing up of the first General-Purpose AI Code of Practice under the EU AI Act, providing legal and technical input on risk management, transparency and compliance mechanisms for general-purpose AI systems.
The Code was finalised in 2025 and now shapes how providers of general-purpose AI models meet their obligations across the European Union.
Advocacy and speaking
Rules are made in rooms. We spend a lot of time with them.
We speak regularly on AI governance, liability, and the convergence of AI with distributed systems, in front of regulators, industry, and academic audiences.
WWC26 - Building Accountability in Agentic AI
P2P Financial Systems International Workshop - Astana 2026
AI-BC and Ethics
Workshop on Competition in Virtual Worlds and Generative AI
AI & Blockchain 2021
Working with us
We advise organisations building and deploying autonomous systems, industry bodies shaping standards, and institutions writing the rules.
The work ranges from legal and regulatory analysis, to drafting, to convening the people who need to be in the same room.
If you are working on any of this, we would like to hear from you.