Building an independent AI Safety Institute focused on alignment in real-world deployments and emerging risks
Providing full-time founder runway to launch NICER: a European institute building the regulatory infrastructure to audit frontier AI agents post-deployment
Providing full-time founder runway to launch NICER: a European institute building the regulatory infrastructure to audit frontier AI agents post-deployment
Project Details
Updated 07/27/26 · Edited by orgNICER (New Intelligence Centre for Emerging Risks from AI) will be an independent AI safety institute based in Central Europe – Warsaw. In March, I transitioned full-time to build this needed organization. My work, which I invested in early research and expert interviews, resulted in the definition of our Institute's vision and the way forward. Right now to realize it I need financial support to focus on this initiative fully.
At NICER, we fix the gap between technical AI alignment research in frontier labs and public-sector enforcement. Researchers build safety mechanisms, but governments lack the engineering capacity to enforce them in the real world. We solve this through training and deployment of "Applied Research Auditors" – researchers who can run behavioural evaluations on autonomous agents and at the same time draft enforceable regulatory standards for European institutions.
Funding from this project would cover my 6-8 month personal runway. I am handling operations, fundraising, and coalition-building so my team of volunteers – including Jan Dubiński, PhD (Astra Fellow at Constellation, Researcher at Truthful AI) and Bart Kubiak (Oxford MPP, AI Policy Officer) – can focus entirely on alignment research.
Without this runway, I will be forced to take corporate consulting contracts, delaying NICER’s launch by 12-18 months and completely missing the critical regulatory window that is open right now.
The concrete outputs for this grant phase are:
-
Formally incorporating the foundation and establishing its legal structures.
-
Signing our first agreements (MoUs) with public institutions in continental Europe to run pilot audit programs.
-
Closing our
$2.5M–3Minstitutional seed round to hire the founding technical team full-time and launch the Fellowship program to train the AI Auditors.
Theory of Impact
Updated 07/27/26 · By grantmaking.aiTechnical alignment research alone won't prevent catastrophic outcomes. If researchers make an alignment breakthrough, it only works if public institutions can deploy and enforce those safety regulations in the real world.
As models reach the recursive self-improvement stage and agentic capabilities, they will interact continuously with real-world infrastructure (national databases, live APIs, shared workspaces). Unsafe behaviours, deployment misgeneralization, and alignment drift will happen post-deployment. Right now, regulators rely on static, point-in-time audits. They completely lack the engineering capacity to conduct continuous behavioural monitoring.
This creates a structural x-risk – governments risk handing critical infrastructure to self-improving agents while completely lacking the technical ability to detect deceptive alignment or loss of control. We want to reduce this critical risk by building the downstream regulatory muscle. We will deploy technical experts who understand both neural networks and regulatory constraints to monitor these systems.
Europe provides a critical strategic wedge via the Brussels Effect. The EU AI Act is the world’s first major AI regulatory framework. If the EU enforces it using static, outdated methodologies, it sets a dangerous global precedent. By solving the government's immediate problem of enforcing the AI Act – NICER architects the underlying technical infrastructure.
People
Updated 07/27/26 · By grantmaking.aiTeam Member
Funding Details
- Mar 1, 2026
- -
- -
- -
- -
- -
- -
- -
- Seeking first grant
- -
Track Record
- Launched AI for Changemakers Accelerator with Google.org, AWS and Lenovo to train 100 executive teams from Social Startups in the deployment of safe AI products. Raised over $4M for that.
- I co-founded Tech To The Rescue – a global platform connecting technology and impact. I scaled it from zero to 35 FTE, raised $6M for it and redeployed $15M+ in tech infrastructure (services, products, compute, credits) for social startups, nonprofits and charities in over 140 countries.
- Finished BlueDot's AGI Strategy and Frontier AI Governance courses.
- Obama Foundation Leader, Member of the Atlantic Council.
Discussion
No comments yet. Be the first to share your thoughts.