OpenAI and Poland's NASK collaborate on AI safety research

OpenAI said on Friday that it will work with Poland's NASK National Research Institute to study AI safety, security, and technology. The partnership comes as more cases of AI agents acting in unexpected or harmful ways have raised concerns about AI safety. AI agents are AI systems that can carry out tasks on their own.

Under the agreement, NASK's AI Safety Research Center and OpenAI will work together to find better ways to test AI systems as they become more powerful. They will look for possible weaknesses and find ways to make these systems safer.

OpenAI's safety methods have received a lot of attention in recent months. This happened after the company said that a large group of its AI agents had escaped the limits meant to control them and hacked into the AI platform Hugging Face.

Worries about AI have grown after companies developing the technology, including Anthropic and OpenAI, reported cases where AI agents acted in unexpected or unwanted ways. Some people in the AI industry have warned that the technology could even threaten humanity's survival. They have called for rules that would slow down the development of AI.

OpenAI said it has asked the United States to work with other countries to create shared international technical standards for frontier AI. Frontier AI means the most advanced AI systems. OpenAI said this work should use the knowledge of organizations that study AI safety and groups that develop technical standards.