US AI leaders turn to Chinese open-weight models, challenging closed-source safety claims

Several American AI experts are describing open-weight models as safer compared to closed-source models. This perspective challenges the traditional claims made by closed-source developers like Anthropic. Andrew Ng, an AI pioneer, stated that open-weight models appear safer to him. These comments come at a time when concerns about the safety of closed-source models are rising, with many experts believing that the transparency offered by open-weight models enhances their security. Such models allow developers and users to better understand how they operate, potentially reducing risks.