US AI Security Order: No Mandatory Model Tests, per Bloomberg

The U.S. government is reportedly planning a new executive order on AI security that won't mandate safety tests for AI models. This news comes from Bloomberg. The order aims to manage the risks of advanced artificial intelligence, like those from companies such as OpenAI and Google. However, it seems to be sidestepping a key recommendation from AI experts: independent pre-deployment testing. This means AI developers might not need to prove their models are safe before releasing them to the public. The bottom line: this decision could significantly impact how quickly and safely new AI technologies reach us.