OpenAI executives said the company is in talks with the Australian side to explore the possibility of training big models of artificial intelligence locally. The move marks an American AI company considering shifting some of its cutting-edge model training work to the Asia-Pacific region. An OpenAI official revealed that the relevant negotiations are still in the negotiation stage, and no final agreement has yet been finalized. The core topics of the negotiations include computing power supply, power security, copyright regulations and regulatory requirements. The Australian Government has been actively lobbying AI giants such as OpenAI, Anthropic, and Nvidia, hoping to attract companies to launch model training services locally and create a sovereign AI computing power ecosystem. However, one of the core obstacles is Australia's current copyright law: current local regulations do not clearly exempt the AI text data training model, and AI companies are worried about facing the risk of copyright disputes. OpenAI stressed that even if the negotiations are successfully implemented, the training program in Australia will advance step by step. Priority will be given to fine-tuning some models, and will not necessarily directly implement the complete pre-training of the largest cutting-edge basic model.

OpenAI executives said the company is in talks with the Australian side to explore the possibility of training big models of artificial intelligence locally. The move marks an American AI company considering shifting some of its cutting-edge model training work to the Asia-Pacific region. An OpenAI official revealed that the relevant negotiations are still in the negotiation stage, and no final agreement has yet been finalized. The core topics of the negotiations include computing power supply, power security, copyright regulations and regulatory requirements. The Australian Government has been actively lobbying AI giants such as OpenAI, Anthropic, and Nvidia, hoping to attract companies to launch model training services locally and create a sovereign AI computing power ecosystem. However, one of the core obstacles is Australia's current copyright law: current local regulations do not clearly exempt the AI text data training model, and AI companies are worried about facing the risk of copyright disputes. OpenAI stressed that even if the negotiations are successfully implemented, the training program in Australia will advance step by step. Priority will be given to fine-tuning some models, and will not necessarily directly implement the complete pre-training of the largest cutting-edge basic model.