News
NASK and CERT Polska Gain Access to OpenAI's Specialized GPT-5.5-Cyber Model
Poland is the only country in the region, and one of the few in the world, to receive access to GPT-5.5-Cyber, a version of OpenAI's model built for cybersecurity teams, as attacks on Polish infrastructure grow.
NASK (Poland's national research institute for cybersecurity) and CERT Polska have been granted access to GPT-5.5-Cyber, a special version of OpenAI's model designed for cybersecurity teams. It marks the first time an institution from Central and Eastern Europe has received this kind of access, and one of the few cases in which the tool has been made available outside the G7 nations.
The Daybreak Program
Daybreak is an OpenAI initiative aimed at reversing the balance of power online in favor of so-called defenders, the blue teams responsible for detecting and neutralizing attacks. Under the Trusted Access for Cyber program, the company gives trusted government partners a version of the model stripped of some of the standard restrictions that would normally block analysis of malicious code or attack techniques.
GPT-5.5-Cyber differs from the standard version of the model through an extended context window and longer working memory, allowing it to analyze much larger chunks of code at once. The tool went to CERT Polska, the team responsible for responding to incidents in Poland's cyberspace, free of charge.
Why Poland
According to NASK, OpenAI's decision stems from the reputation the institute and CERT Polska have built in international technical circles through years of rigorous analytical work. NASK director Dr. Radosław Nielek stresses that this recognition places Poland among the countries the American company treats as trusted partners in digital security.
This is genuinely an elite group, since we are one of only a few countries in the world, alongside the G7 nations, and the only one in the region with this kind of access - Dr. Radosław Nielek, director of NASK
The model is integrated into CERT Polska's agent-based systems through an API. Reports generated from the analyses are then aggregated by the model itself, which is meant to shorten the time needed to go from detecting a vulnerability to preparing recommendations for public administration and critical infrastructure operators.
A Race Against Time
Nielek points out that the same technology speeding up the work of defenders also makes life easier for attackers. In his view, criminals today can analyze a freshly published security patch and build a working exploit from it in hours rather than the weeks it used to take.
This model is what's known as a force multiplier. It means outstanding specialists get much more done, so they become more effective and work faster - Dr. Radosław Nielek, director of NASK
This race is unfolding against a broader geopolitical backdrop. NASK links the December 2025 attack, which targeted wind farms and a combined heat and power plant among other things, to growing activity by Russian groups aimed at Poland's energy infrastructure. The institute's director describes Poland as a frontline state, noting that proximity to the ongoing war directly translates into the number and intensity of attacks in cyberspace.
Humans Still Decide
Despite his enthusiasm for the new tool, the NASK director stresses that the model does not replace analysts. In his assessment, artificial intelligence accounts for at most one-fifth of the success in detecting and neutralizing threats, while the knowledge and experience of the human specialists working at CERT Polska remains decisive.
For Polish companies and public institutions, this means a real change in the pace of responding to reported vulnerabilities, especially within the administration, where patching gaps had until now often stretched over weeks. NASK says it will also use its access to the model when analyzing reports coming in from critical infrastructure operators and smaller entities that lack their own security teams.
The institute simultaneously runs its own on-premise AI models, built for working with sensitive data that, for security reasons, cannot be sent to external providers. GPT-5.5-Cyber is meant to complement this toolkit in cases where scale and speed of analysis matter more than the confidentiality of specific data.
Sources: OpenAI GPT-5.5-Cyber in Poland, interview with Dr. Radosław Nielek (android.com.pl), Head of NASK: we are the only country in the region with access to this OpenAI technology (rp.pl), Poland in an elite group as NASK gains access to GPT-5.5 Cyber from OpenAI (itwiz.pl)
