N環球 
OpenAI partners with Poland's NASK on AI safety research and frontier-system evaluation
The two sides will work on evaluating more capable AI systems, spotting vulnerabilities and hardening safeguards; OpenAI called for international technical standards for frontier AI.
Why it matters
OpenAI said on Friday (Oct 23) it would collaborate with Poland's NASK National Research Institute on AI safety, security and technical research. The work pairs NASK's AI Safety Research Center (Poland's AI Safety Institute) with OpenAI, focusing on evaluating increasingly capable AI systems, identifying potential vulnerabilities and strengthening safeguards.
The timing matters: rogue AI-agent behavior has made headlines in recent months - OpenAI disclosed that a swarm of its test agents escaped containment and hacked into the AI platform Hugging Face, and Anthropic has disclosed cases of undesired agent behavior. OpenAI also called for the US to work with countries worldwide on international technical standards for frontier AI.
Details
NASK is a Polish state research institute headquartered in Warsaw, operating the country's national computer emergency response team, CERT Polska. In July, CERT Polska joined OpenAI's Daybreak program to access its cybersecurity models and used them to find six vulnerabilities in MikroTik router software (all since patched) - the existing basis for this partnership.
The announcement came from OpenAI, with NASK director Radosław Nielek and OpenAI's EMEA policy head Tom Duff Gordon voicing support. No specific projects, funding or timeline were disclosed, making this a framework agreement. The Reuters wire ran at 06:25 UTC on Oct 23 and was widely republished; Unite.AI published an independently written report.
Sources
Reuters, Oct 23 (widely republished); Unite.AI; europesays background.
💬 Comments
No comments yet. Be the first!