OpenAI signs agreement with Poland’s NASK to research AI risks and safeguards

OpenAI and Poland’s NASK National Research Institute signed an agreement announced October 23, 2026, to explore joint research on AI safety, security and technical questions. The partnership brings Poland’s AI safety institute into work aimed at evaluating more capable systems, identifying potential weaknesses and strengthening safeguards—not a completed assessment of whether those systems are safe.

The relationship began with cyber defense

The agreement follows a narrower collaboration earlier this year. In July, CERT Polska, which operates within NASK, gained access to OpenAI’s cybersecurity-capable models through its Daybreak program. NASK is a state-owned research institute based in Warsaw; CERT Polska is the national team responsible for responding to cyberattacks on Polish networks.

OpenAI says the team used its models to help discover six vulnerabilities in MikroTik RouterOS, the software running MikroTik routers. Some flaws could be combined to take control of affected devices without authentication. The vendor released fixes, and CERT Polska confirmed that those fixes prevented the attacks it observed, according to OpenAI’s announcement.

That earlier work concerned using AI to find weaknesses in other software.

A broader research agenda, not a safety verdict

NASK’s AI Safety Research Center will take part in the collaboration. The center serves as Poland’s Artificial Intelligence Safety Institute, giving the agreement a public-research counterpart to OpenAI’s own technical teams.

NASK director Radosław Nielek described the center’s existing work as spanning safeguards against emerging risks, detection of AI-generated content, malicious uses of that content and challenges across languages.

For OpenAI, Tom Duff Gordon, its vice president and head of EMEA policy, framed the agreement as a way to bring in expertise beyond the company. His stated aim includes having outside researchers challenge OpenAI’s assumptions, test its systems and help strengthen safeguards.

AI safety cannot be achieved in isolation. It is a collaborative effort, and only through active cooperation between public AI Safety Institutes and frontier AI developers can we ensure that AI becomes truly reliable and safe to use.

Radosław Nielek, director of NASK, in OpenAI’s announcement

National expertise meets international standards

OpenAI places the agreement alongside its existing safety work with the UK AI Security Institute and the US Center for Advancing Standards and Innovation for Super Intelligence, known as CAISSI. The company says collaboration with public research institutions has helped strengthen safeguards and identify and fix security vulnerabilities in AI agents.

It is also advocating international technical standards for frontier AI—the most advanced systems. OpenAI’s argument is that common approaches to evaluating capabilities, assessing risks, testing safeguards and reporting incidents would let countries compare evidence and respond to risks that cross borders.

The company describes NASK’s contribution as grounded in protecting public services and Poland’s digital information environment. It argues that national partners can keep global standards connected to how AI is actually used.