OpenAI and NASK Agree to Research AI Safety

Conceptual cover illustration.

On October 23, OpenAI announced a signed agreement with Poland’s NASK institute to explore collaboration on artificial intelligence safety and security, as well as technical research. The announced work includes studying how to evaluate increasingly capable AI systems, identify vulnerabilities, and strengthen safeguards. NASK is participating through its AI Safety Research Center, which serves as the Polish Artificial Intelligence Safety Institute (AISI). OpenAI describes the agreement and its scope.

Areas of work and content safety

NASK Director Radosław Nielek describes the institute’s areas of work, including safeguards against emerging risks, detection of AI-generated synthetic content, and analysis of malicious uses. He also mentions multilingual challenges. These topics connect safety research with questions of authenticity and content across different languages, without announcing specific tools for verifying posts. The announcement outlines the areas Nielek identifies.

At OpenAI, Tom Duff Gordon is vice president and head of policy for EMEA. The partnership is framed as an exploratory collaboration between the two organizations. OpenAI places it within an international effort that includes comparing methods for evaluating capabilities, assessing risks, testing safeguards, and reporting incidents. The stated goal is to contribute to responses to risks that can cross borders. OpenAI presents the collaboration and its international context.