Poland honored by the USA. NASK will use the groundbreaking Open AI tool

This is a breakthrough for Polish cybersecurity and honors the work of experts from CERT Polska, operating within NASK. The institute’s specialists work every day on how language models can be used even better and more effectively to increase security in cyberspace, and thanks to access to GPT5.5-cyber, these possibilities will be even greater.

GPT5.5 cyber will speed up security analysis

Thanks to the GPT5.5-cyber model, the work of cybersecurity analysts will speed up and become much easier, mainly in searching for vulnerabilities and analyzing malware.

– Most off-the-shelf LLMs refuse to perform most of the tasks that the cyber model does. Why? This information can be used for malicious purposes, e.g. to carry out a cyber attack.

– The GPT5.5-cyber model is trained to look for vulnerabilities. He will not refuse to execute such a command, but for this reason it can only be made available to trusted institutions. Such as NASK – explains Radosław Nielek, director of NASK.

The GPT5.5-cyber model is a specialized solution for cybersecurity guards. The White House gave the “green light” to make this system available to Poland, thanks to which we have become one of the few countries in the world that can use the latest OpenAI tool intended for cybersecurity tasks.

GPT-5.5 can not only be a shield to detect vulnerabilities in our own security, but also find weaknesses in hostile hackers, called vulnerabilities.

– By providing us with GPT Cyber, OpenAI guarantees that we will use it only for the right purpose, i.e. to look for vulnerabilities or counteract cybercriminals – says Radosław Nielek.

OpenAI tool designed for cybersecurity tasks

The GPT5.5 model from Open AI is made available only to trusted partners. Experts will gain advanced access through the Trusted Access for Cyber ​​program for government entities.

– CERT Polska experts prove that they have the competence and knowledge to take full advantage of these opportunities. Previous research in the area of ​​using language models to improve cybersecurity has certainly contributed to NASK gaining such an opportunity as part of the Daybreak program – explains Radosław Nielek.

The decision is part of Daybreak, an initiative by Sam Altman

Daybreak is an initiative that aims to keep defenders one step ahead of criminals in the use of artificial intelligence in cybersecurity. It was created by Sam Altman so that in the Internet – in the age of AI – those who defend us with attacks, rather than the attackers, would gain more power.

GPT-5.5-Cyber ​​will be used to analyze malware, perform computer forensics tasks, or find vulnerabilities. The focus is on accelerating the entire vulnerability detection and remediation cycle with leading global cybersecurity organizations as partners. The creators of the solution emphasize that these tasks should be carried out responsibly.

OpenAI has been systematically developing its models since 2015, strengthening its position as one of the global leaders in this area.

AI has changed cybersecurity

Language models significantly speed up the detection of vulnerabilities, which has been especially noticeable in recent months. Previously, finding vulnerabilities was the most complex process because it required rare knowledge, time and deep knowledge of complex systems. Currently, models can navigate large code bases and analyze possible attack paths, all in a much shorter time. Defenders are overwhelmed by the number of vulnerabilities they find, and remediating them becomes crucial.

– Our experts absolutely need access to these solutions to efficiently detect and repair vulnerabilities before attackers do it – says Marcin Dudek, head of CERT Polska, adding that there are many possibilities of using new tools. GPT-5.5-Cyber ​​will be used for malware analysis, computer forensics tasks and searching for vulnerabilities. We already share our knowledge of using large language models on our website and plan to continue doing so. All these elements significantly contribute to building a safer cyberspace.