Skip to content
ReviewRamble

The weekly brief

AI, reviewed, in your inbox.

  • New reviews of tools, models, MCP servers and repos, with the verdict up front
  • Re-checked prices and ratings when something changes
  • New analysis pieces as they are published

Models, agents, MCP servers and the tools actually worth your time.

One email a week. Unsubscribe with the link in any brief. How we use your email

AI news & research

News ·

OpenAI signs an AI safety research agreement with Poland's NASK

OpenAI and NASK, the Polish research institute whose AI Safety Research Center acts as Poland's AI Safety Institute, will work together on evaluating capable models, finding vulnerabilities and strengthening safeguards. The deal builds on CERT Polska, part of NASK, gaining access to OpenAI's cyber-capable models through Daybreak in July. OpenAI already works with the US CAISSI and the UK AISI.

OpenAI has signed an agreement with NASK, Poland's National Research Institute, to explore joint work on AI safety, security and technical research. NASK's AI Safety Research Center acts as Poland's AI Safety Institute, so the deal adds another national body to the list of public evaluators OpenAI works with, alongside the US Center for Advancing Standards and Innovation for Super Intelligence (CAISSI) and the UK AI Security Institute.

The stated aims are broad: finding ways to evaluate increasingly capable systems, identifying vulnerabilities and strengthening safeguards. The more concrete part is what came before. In July, CERT Polska, the national computer emergency team that sits inside NASK, gained access to OpenAI's cyber-capable models through its Daybreak programme. OpenAI says the team used them to help find six vulnerabilities in MikroTik RouterOS, the software on a widely used family of routers, including flaws that could be chained to take over devices without logging in. MikroTik shipped fixes, and CERT Polska confirmed they stopped the attacks it had seen.

OpenAI frames the agreement as part of a push for international technical standards on evaluating capabilities, testing safeguards and reporting incidents, built with the help of national safety institutes. That fits a pattern in our recent coverage: Anthropic giving METR deep access to its models, the UK institute's tests of Claude Mythos 5, and OpenAI's own proposals for cyber defence. Labs increasingly want public bodies inside the evaluation process, partly because governments are asking for it.

The announcement is short on detail. It says nothing about which models NASK will be able to test, how early it will see them, whether it can publish its findings, or how disagreements would be handled. It is described as an agreement to explore collaboration, not a programme with deliverables. The RouterOS work shows the cyber side can produce real fixes; whether the safety research yields public evidence, rather than private reassurance, is the test to watch.

Our take: Agreements with national safety institutes are becoming routine, and this one is light on specifics such as model access or publication terms. Its value depends on whether NASK can publish what it finds.

Source: OpenAI

More stories