OpenAI and Poland's NASK Partner on AI Safety and Security
OpenAI and Poland's NASK National Research Institute have signed an agreement to explore collaboration on AI safety, security and technical research. The partnership matters because it adds a national public research institution to the emerging network of government-backed AI safety institutes that OpenAI already works with, and because NASK's CERT Polska has already used OpenAI's cyber capable models to find real vulnerabilities in widely deployed networking hardware.
What the agreement covers
The agreement is a framework for collaboration rather than a product release. Under it, NASK's AI Safety Research Center — which serves as the Polish Artificial Intelligence Safety Institute (AISI) — and OpenAI will work together on AI safety research.
The stated areas of joint work are:
- Exploring ways to evaluate increasingly capable AI systems.
- Identifying potential vulnerabilities in those systems.
- Strengthening safeguards.
The two organizations describe the goal as bringing together expertise from both sides to better understand emerging risks, strengthen safeguards, and explore how advanced AI can contribute to a more secure and resilient digital ecosystem.
Why OpenAI frames this as an international effort
OpenAI states that as AI systems become more capable, international collaboration between developers and public research institutions will be essential to understanding their risks and building stronger safeguards.
The company points to its existing work with public research institutions, including the US Center for Advancing Standards and Innovation for Super Intelligence (CAISSI) and the UK AI Security Institute (AISI). According to OpenAI, that work has already helped strengthen safeguards, identify and fix security vulnerabilities in AI agents, and advance frontier AI safety.
OpenAI has separately called for the US to work with countries around the world on international technical standards for frontier AI, drawing on the expertise of AI safety institutes and standards bodies. The company argues that common approaches to evaluating capabilities, assessing risks, testing safeguards and reporting incidents can help countries compare evidence and respond to risks that cross borders. The NASK collaboration is presented as an example of bringing national expertise into that wider effort.
NASK's role and the Polish context
NASK's contribution rests on extensive research in AI safety that is applied directly to protecting public services and Poland's digital information ecosystem. OpenAI's post makes a specific argument about why national partners matter: without national partners like NASK, global AI safety standards may remain detached from the realities in which AI is actually used.
That framing positions AI safety institutes as a bridge between abstract international standards and the concrete environments — public services, national digital infrastructure — where AI systems are deployed.
Prior work: CERT Polska and the MikroTik RouterOS vulnerabilities
The agreement builds on existing cooperation between OpenAI and cyber defenders in Poland. In July, CERT Polska, which operates within NASK, gained access to OpenAI's cyber capable models through Daybreak.
According to the announcement, the team used OpenAI models to help discover six vulnerabilities in MikroTik RouterOS, including flaws that could be combined to take control of affected devices without authentication. The vendor released fixes, which CERT Polska confirmed prevented the attacks it observed.
This is the concrete precedent cited for the partnership: a national computer emergency response team using frontier models to find and help remediate real security flaws in networking equipment, with the vendor shipping patches and the fixes being verified against observed attacks.
What the announcement does and does not say
The post describes an agreement to explore collaboration and lists the intended research areas. It does not specify funding, staffing, timelines, deliverables, or any joint benchmarks or evaluation results. The only quantified outcome cited is the earlier CERT Polska work: six vulnerabilities found in MikroTik RouterOS, with vendor fixes confirmed to have prevented the attacks CERT Polska observed.