OpenAI and Poland’s NASK National Research Institute signed an agreement to explore joint research into artificial intelligence safety, security and safeguards for increasingly capable systems, OpenAI said on Oct. 23.
The agreement extends a relationship with Polish cyber defenders that helped identify software flaws earlier this year, the company said. OpenAI, an AI developer, announced it in “OpenAI and Poland’s NASK partner to advance AI safety and security,” published on Oct. 23.
NASK’s AI Safety Research Center, which serves as Poland’s Artificial Intelligence Safety Institute, will work with OpenAI to explore ways to evaluate more capable systems, the company said. Their planned research also covers finding potential vulnerabilities and strengthening safeguards.
The agreement brings together the organizations’ expertise to investigate emerging risks and explore how advanced AI could help make digital systems more secure, OpenAI said. It described the arrangement as an agreement to explore collaboration.
The partnership builds on work involving CERT Polska, a cybersecurity team within NASK, which received access to OpenAI’s models with cybersecurity capabilities through Daybreak in July, the company said.
CERT Polska used those models to help discover six vulnerabilities in MikroTik RouterOS, OpenAI said. Some of the flaws could be used together to take control of affected devices without authentication, the process used to verify access.
The vendor released fixes, and CERT Polska confirmed that those fixes prevented the attacks the team had seen, the company said. OpenAI linked that earlier cybersecurity work to the broader research relationship established by the agreement.
OpenAI said NASK brings experience from AI safety research applied to protecting public services and Poland’s digital information environment. It said national partners help connect international safety standards with the conditions in which people actually use AI.
The company said it has called for the United States to work with other countries on international technical standards for the most advanced AI systems. That effort would draw on safety institutes and organizations that develop standards.
Shared methods for evaluating capabilities, assessing risks, testing safeguards and reporting incidents could help countries compare evidence and address risks across borders, OpenAI said. It placed the NASK agreement within that wider international effort.
OpenAI said its existing work with public research institutions includes the US Center for Advancing Standards and Innovation for Super Intelligence and the UK AI Security Institute. Those relationships have helped strengthen safeguards and find and fix security vulnerabilities in AI agents, the company said.
