OpenAI and Poland’s NASK have signed an agreement to explore joint work on AI safety, security, and technical research, giving Polish public researchers access to frontier models and the testing workflows that surround them.
NASK’s AI Safety Research Center – which serves as the Polish Artificial Intelligence Safety Institute (AISI) – will work with OpenAI on evaluating increasingly capable systems, identifying potential vulnerabilities, and strengthening safeguards.
The clearest operational example came before the signing. In July, CERT Polska, the incident response team that operates within NASK, gained access to OpenAI cyber-capable models through a channel named Daybreak. Analysts pointed those models at MikroTik RouterOS and found six vulnerabilities, including flaws that could be chained to take control of affected devices without authentication.
MikroTik released fixes for the RouterOS flaws. CERT Polska then confirmed the patches prevented the attacks its analysts had recorded against unpatched hardware. Model access fed vulnerability discovery, vendor remediation followed, and validation ran against the original attack paths.
That testing loop is what OpenAI wants to repeat with national institutes. Poland isn’t a newcomer to the company’s research effort either; OpenAI notes that many of the scientists who have helped lead its own work are Polish.
NASK applies its safety research to protecting public services and Poland’s digital information environment. The centre’s remit includes detecting synthetic AI-generated content and tracking how such material is used maliciously across different languages. That multilingual angle carries weight in a country whose public messaging isn’t confined to a single language.
Radosław Nielek, Director of NASK, said: “Collaboration with OpenAI is a natural step for the Institute and its AI Safety Research Center, which is actively involved in AI safety research. This work spans developing safeguards against emerging risks, detecting synthetic AI-generated content, and understanding both its malicious uses and the multilingual aspects of these challenges.
“AI safety can’t 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.”
Tom Duff Gordon, VP and Head of EMEA Policy at OpenAI, added: “As AI becomes more capable, it’s vital that we draw on expertise beyond our own organisation to challenge our assumptions, test our systems, and strengthen safeguards.
“This agreement with NASK opens the door to doing that together, bringing Poland’s research expertise into the international effort to make AI safer.”
OpenAI points to its existing work with the US Center for Advancing Standards and Innovation for Super Intelligence (CAISSI) and the UK AI Security Institute ( AISI ), which it says has already helped identify and fix security vulnerabilities in AI agents.
The company has called for US coordination with other countries on international technical standards for frontier AI, drawing on safety institutes and standards bodies.
OpenAI sets out four shared activities for national institutes. Capability evaluation and risk assessment come first. Safeguard testing and incident reporting follow, so countries can compare evidence when a risk crosses a border.
The testing cycle ended with vendor patches in place. CERT Polska confirmed the fixes prevented the attacks its analysts had recorded, covering all six RouterOS vulnerabilities.
See also: State legislatures become de facto AI regulators as Congress stalls
Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events including IoT Tech Expo and the Cyber Security & Cloud Expo . Click here for more information.
AI News is powered by TechForge Media . Explore other upcoming enterprise technology events and webinars here .
