OpenAI signs AI safety agreement with Poland's NASK research institute

The agreement involves NASK's AI Safety Research Center, which serves as the Polish Artificial Intelligence Safety Institute (AISI). The organizations will examine methods for evaluating increasingly capable AI systems, identifying security vulnerabilities and strengthening safeguards against emerging threats. The collaboration builds on existing cybersecurity work between OpenAI and Polish researchers, including a project in which CERT Polska, the country's computer emergency response team operating within NASK, used OpenAI models to uncover six vulnerabilities in MikroTik RouterOS software.

George Osborne, Head of OpenAI for Countries, confirmed the agreement in a LinkedIn post, highlighting NASK's experience in protecting public infrastructure.

"NASK brings experience protecting public services, helping ground international safety research in the challenges institutions face in practice." OpenAI and NASK to examine AI risks and security safeguards The agreement establishes a framework for research into the safety and security of advanced AI systems, with particular attention to how potential risks can be identified and evaluated.

NASK brings experience in cybersecurity research, including developing safeguards against emerging threats, detecting AI-generated synthetic content and investigating the malicious use of AI. Its work also covers multilingual security challenges, an area where risks may differ depending on the languages and digital environments in which AI systems operate. The institute conducts research across cybersecurity, artificial intelligence and cloud computing, with three research divisions and 14 specialized departments.

Alongside its research activities, NASK supports Poland's public sector through digital infrastructure projects, cybersecurity services and training. Its work includes connecting schools to the internet, developing secure document management systems for government agencies and providing cybersecurity training for public officials. Radosław Nielek, Director of NASK, says: "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." The agreement is exploratory, with the organizations identifying areas for potential technical collaboration rather than announcing the completion of a specific joint research project. Previous collaboration identified six RouterOS vulnerabilities The new agreement follows earlier cooperation involving CERT Polska and OpenAI's cybersecurity tools. In July, CERT Polska gained access to OpenAI's cyber-capable models through Daybreak, an initiative providing access to advanced AI capabilities for cybersecurity work.

Using the models, researchers identified six vulnerabilities in MikroTik RouterOS, software used to manage networking devices. According to OpenAI, some of the vulnerabilities could be combined to enable attackers to take control of affected devices without authentication. MikroTik subsequently released fixes for the identified flaws.

The case provides a concrete example of how the organizations have already worked together on cybersecurity, while the new agreement broadens their potential cooperation to the evaluation and security of advanced AI systems themselves. International collaboration on AI safety standards OpenAI's partnership with NASK also extends its existing work with public AI safety research organizations, including the UK's AI Security Institute and the US Center for Advancing Standards and Innovation for Super Intelligence (CAISSI). The company has been advocating international technical standards for frontier AI, with shared approaches to evaluating model capabilities, assessing risks, testing safeguards and reporting security incidents.

Such frameworks are intended to help researchers and public authorities compare findings across different organizations and national jurisdictions. Tom Duff Gordon, VP and Head of EMEA Policy at OpenAI, says: "As AI becomes more capable, it’s vital that we draw on expertise beyond our own organization to challenge our assumptions, test our systems and strengthen safeguards." The OpenAI-NASK agreement will explore how Poland's national research expertise can contribute to that work, including assessing increasingly capable AI systems and developing methods to identify security weaknesses before they are exploited. ETIH Innovation Awards 2027 open for entries The ETIH Innovation Awards 2027 are now open for entries, with 26 categories recognising companies, institutions, projects and individuals making a measurable impact across education technology.

The global awards span AI in education, digital learning, higher education, skills and workforce, inclusion, education operations, evidence and impact, with entries managed through ETIH’s new online awards platform. Entries are open worldwide, with the early-bird deadline on Friday 18 December 2026 and the final entry deadline on Friday 26 February 2027 at 23:59 UK time. Explore the 2027 categories and judging criteria, read the How to Enter guide or start an entry through the ETIH Innovation Awards 2027 platform.