OpenAI and NASK, Poland's National Research Institute, signed an agreement on 23 October to collaborate on AI safety, security and technical research. NASK's AI Safety Research Center serves as the Polish AI Safety Institute, and its director Radosław Nielek describes its remit as safeguards against emerging risks, detecting AI-generated content, studying malicious uses of AI and the multilingual dimension of all three. The stated work is evaluating increasingly capable systems, identifying vulnerabilities and strengthening safeguards. No funding, duration or deliverables were published.
What distinguishes this from the usual memorandum is that the working relationship predates it. CERT Polska, which operates within NASK, received access to OpenAI's cyber-capable models in July through the Daybreak programme. Using them, its team helped discover six vulnerabilities in MikroTik RouterOS, some of which could be chained to take control of affected devices without authentication. The vendor shipped fixes and CERT Polska confirmed they prevented the attacks it had observed.
Daybreak is the same programme Fortinet credited this week for the critical flaws it patched across five product lines on 21 and 22 October, alongside Anthropic's Glasswing. Two national CERTs and one major vendor now point at the same two programmes, which makes frontier-model vulnerability discovery a working pipeline rather than a demonstration. OpenAI also lists the US CAISSI and the UK AISI as existing collaborators, though neither is part of this agreement.
Sources
OpenAI · TNW · Unite.AI