OpenAI Teams Up with Poland's NASK on AI Safety and Security

OpenAI and Poland's NASK National Research Institute signed an agreement to explore collaboration on artificial intelligence safety, security, and technical research (OpenAI). The partnership combines expertise from both groups to assess emerging risks, tighten safeguards, and examine how advanced AI can support a secure digital ecosystem.

Under the pact, NASK's AI Safety Research Center—Poland's Artificial Intelligence Safety Institute—will work with OpenAI to evaluate increasingly capable AI systems. Both sides aim to spot potential vulnerabilities and reinforce safeguards through joint technical work.

The deal builds on earlier cooperation between OpenAI and Polish cyber defenders. In July, CERT Polska, part of NASK, accessed OpenAI's cyber‑capable models via Daybreak.

During that testing phase, CERT Polska used the models to uncover six vulnerabilities in MikroTik RouterOS. The flaws could be chained to seize control of affected devices without authentication.

After the findings, the vendor issued fixes for the RouterOS issues. CERT Polska later confirmed the patches blocked the attacks it had observed.

OpenAI's work with NASK highlights the benefit of weaving national expertise into broader international AI safety efforts. The company has previously urged the United States to engage globally on technical standards, drawing on AI safety institutes and standards bodies.

Without partners like NASK, global AI safety standards risk staying detached from the real‑world environments where AI is deployed. International cooperation lets countries compare evidence and address cross‑border risks.

The partnership between OpenAI and Poland's NASK suggests a growing trend of integrating national research institutes directly into frontier AI development and security frameworks. By formalizing cooperation with the Polish Artificial Intelligence Safety Institute, OpenAI appears to be bridging the gap between high-level safety research and regional operational defense, as demonstrated by CERT Polska's previous vulnerability discoveries.

Spotted an error? See our corrections policy or email editorial@lordofclaude.com.