OpenAI says GPT-4 poses "at most" a slight risk of helping create biological threats, according to early tests to evaluate "catastrophic" risks from LLMs (Rachel Metz/Bloomberg)

Rachel Metz / Bloomberg:

OpenAI says GPT-4 poses “at most” a slight risk of helping create biological threats, according to early tests to evaluate “catastrophic” risks from LLMs  —  Artificial intelligence startup carried out tests as part of efforts to understand and prevent any “catastrophic” risks from its technology.