OpenAI and NASK launch AI safety research collaboration in Poland
OpenAI and Poland’s NASK National Research Institute have signed an agreement to explore collaboration on AI safety, security and technical research. NASK’s AI Safety Research Center, which serves as the Polish Artificial Intelligence Safety Institute (AISI), will work with OpenAI on ways to evaluate increasingly capable AI systems, identify potential vulnerabilities and strengthen safeguards.
The organisations said the partnership combines their expertise to improve understanding of emerging risks and examine how advanced AI can support a more secure and resilient digital ecosystem. The announcement does not specify a timetable, funding arrangement or individual research projects.
Safety research and public-sector expertise
NASK’s work includes safeguards against emerging risks, detection of synthetic AI-generated content, malicious uses of AI and the multilingual dimensions of those challenges. NASK director Radosław Nielek said AI safety requires active cooperation between public AI safety institutes and frontier AI developers.
OpenAI said it needs expertise beyond its own organisation as AI systems become more capable, both to challenge assumptions and to test systems and safeguards. The company cited existing work with the US Center for Advancing Standards and Innovation for Super Intelligence, the UK AI Security Institute and other public research institutions.
That international effort also connects with governance inside the developer: the appointment of Paul Christiano’s OpenAI safety oversight role to the OpenAI Foundation board and safety committee illustrates how safety oversight is being expanded alongside external technical collaboration.
Standards, evaluations and cyber defence
OpenAI has called for the United States to work with other countries on international technical standards for frontier AI. It said common approaches to capability evaluations, risk assessment, safeguard testing and incident reporting could help countries compare evidence and respond to risks that cross borders.
The Polish agreement builds on a separate cyber-security engagement. In July, CERT Polska, which operates within NASK, received access to OpenAI’s cyber-capable models through Daybreak. CERT Polska used the models to help discover six vulnerabilities in MikroTik RouterOS, including flaws that could be combined to take control of affected devices without authentication.
MikroTik released fixes, and CERT Polska confirmed that those fixes prevented the attacks it had observed. The case gives the new research agreement a concrete operational context: it links AI safety research with the protection of public services and Poland’s digital information ecosystem.
What organisations can take from the agreement
For businesses deploying or developing AI, the announcement underlines the value of repeatable evaluations, vulnerability testing and clear incident processes rather than relying only on internal assessments. Working with relevant public research and security institutions can help connect technical safeguards to the environments in which AI is actually used.