Poland Gains Access to GPT-5.5 Cyber, Model Deployed at NASK and CERT Polska

NASK has been granted access to GPT-5.5 Cyber, a specialized version of OpenAI's model built to detect security vulnerabilities and automatically patch code. The Polish institution became the third organization in Europe, and the first in the entire Central and Eastern European region, to receive this kind of access, with the tool set to become part of CERT Polska's day-to-day work.

What the tool does

GPT-5.5 Cyber is a variant of the GPT-5.5 model with an extended context window and working memory, optimized for code analysis and digital forensics. The model can scan massive codebases, flag components relevant to security, check whether a vulnerable piece of code is actually reachable by an attacker, test suspected vulnerabilities in a controlled environment, and prepare fixes for human review.

OpenAI is making it available through its Trusted Access for Cyber program, part of the broader Daybreak initiative focused on securing critical infrastructure worldwide. Vetted companies such as Check Point have already used similar access, and the program also includes government partnerships with Australia, Canada, France, Germany, Japan, South Korea, and EU institutions.

The role of NASK and CERT Polska

NASK director Radosław Nielek stressed that access to the model is no accident, but rather a result of the recognition the institute and its experts have earned worldwide. CERT Polska head Marcin Dudek added that his team needs tools like this to detect and fix vulnerabilities before attackers do.

Speaking for OpenAI, Tom Duff-Gordon, the company's head of policy for Europe, the Middle East and Africa, said he was proud of the collaboration with Polish partners, whose technical expertise is key to strengthening critical infrastructure security.

How much faster

According to NASK, vulnerability searches that used to take analysts weeks can now be completed by the model in hours, as it scans vast public administration code repositories. The institute stresses that final verification of findings and decisions on deploying fixes remain in human hands.

That caveat matters given the rising number of attacks on public systems and the shrinking window defenders have to react once a new vulnerability comes to light. Automating the first stage, finding and preliminarily describing a vulnerability, is meant to relieve analysts, who today are the bottleneck in many security teams.

What it means for Poland

For Poland's cybersecurity sector, this is a signal that domestic institutions are being treated by global AI providers as partners rather than just recipients of finished solutions. Access to the most advanced, distribution-restricted models usually goes first to the US and a handful of allies, so Poland's place in this group also carries symbolic and political weight.

In practice, this also means that public administration bodies and companies working with CERT Polska can expect faster patching of reported vulnerabilities, narrowing the window cybercriminals can exploit. How quickly similar tools reach the private sector and smaller institutions remains an open question.

Sources: NASK Gets a Breakthrough Tool from OpenAI (nask.pl), Poland Gains Access to GPT-5.5 Cyber, Tool Heading to CERT Polska (bankier.pl), NASK: Poland Third Country in Europe with GPT-5.5 Cyber Access (itreseller.pl), Scaling Trusted Access for Cyber with GPT-5.5 and GPT-5.5-Cyber (openai.com)