OpenAI has signed an agreement with Poland's NASK National Research Institute to collaborate on artificial intelligence safety, cybersecurity and technical research, as concerns grow over the potential risks associated with increasingly capable AI systems.
Announced on October 23, the partnership will bring together OpenAI's technical capabilities and NASK's research expertise to evaluate advanced AI systems, identify security vulnerabilities and develop safeguards against emerging threats.
Under the agreement, NASK's AI Safety Research Center, which serves as the Polish Artificial Intelligence Safety Institute (AISI), will work with OpenAI on research examining the capabilities and potential risks of advanced AI models.
The collaboration will focus on improving methods for assessing AI systems, detecting vulnerabilities and strengthening protective measures as AI technologies become more widely deployed across industries and public services.
According to OpenAI, international cooperation between AI developers and public research institutions is becoming increasingly important as advanced models introduce new technical and security challenges.
The company said its existing collaborations with the US Center for Advancing Standards and Innovation for Super Intelligence (CAISSI), the UK AI Security Institute and other research organisations have contributed to identifying vulnerabilities and improving safeguards for AI systems.
Radosław Nielek, Director of NASK, said the institute's research includes developing safeguards against emerging risks, detecting AI-generated synthetic content and examining the malicious use of artificial intelligence.
He also highlighted the importance of addressing multilingual challenges associated with AI safety and strengthening cooperation between public research institutions and developers of advanced models.
Tom Duff Gordon, Vice President and Head of EMEA Policy at OpenAI, said Poland's scientific and technical expertise could contribute to international efforts to evaluate AI systems and improve their reliability.
The agreement builds on earlier cooperation between OpenAI and Polish cybersecurity researchers.
In July, CERT Polska, which operates within NASK, received access to OpenAI's cyber-capable models through the company's Daybreak programme.
Using these models, the cybersecurity team identified six vulnerabilities in MikroTik RouterOS, including weaknesses that could potentially be combined to gain unauthorised control of affected devices.
MikroTik subsequently released security fixes, which CERT Polska confirmed prevented the attacks observed during testing.
The latest partnership also comes as AI developers face increased scrutiny over autonomous agents, cybersecurity vulnerabilities and the potential misuse of advanced models.
OpenAI has advocated international technical standards for frontier AI, including common approaches to evaluating capabilities, assessing risks, testing safeguards and reporting security incidents.
The company said cooperation with national research institutions could help ensure that global AI safety standards reflect the practical challenges faced by governments and organisations deploying these technologies.
Financial terms, specific research milestones and a completion timeline for the partnership have not been disclosed.
Disclaimer: This article may include information derived from interviews, press releases, public statements, research, company communications and other publicly available or third-party sources. Such material may be summarised, paraphrased or contextualised for journalistic and editorial purposes. All rights in third-party content remain with their respective owners.