OpenAI 稱中國競爭對手利用其成果開發 AI 應用

OpenAI, the organization behind the popular AI tool ChatGPT, has raised concerns that some of its competitors, particularly those based in China, are leveraging its innovations to enhance their own artificial intelligence applications. A notable example is DeepSeek, a Chinese application that reportedly mimics the capabilities of ChatGPT but does so at a significantly lower cost. This situation has sparked worries about the competitive landscape of AI technology, as OpenAI and other American companies have long been recognized as frontrunners in this field. Reports indicate that Microsoft, a major investor in OpenAI, is currently investigating whether DeepSeek has improperly utilized data belonging to OpenAI. The BBC has reached out to both Microsoft and DeepSeek for their perspectives on this matter. Recently, David Sacks, who has been appointed as the White House's AI and cryptocurrency advisor, expressed his views during an appearance on Fox News. He suggested that DeepSeek may have employed a technique known as knowledge distillation, which involves extracting valuable insights from OpenAI's models to enhance its own offerings. Sacks stated, 'There’s substantial evidence that what DeepSeek did here is they distilled the knowledge out of OpenAI’s models. ' He anticipates that in the coming months, leading AI companies in the United States will take measures to prevent such practices, which he believes could hinder the development of similar applications. The US government has already initiated steps to safeguard its advancements in AI technology by implementing regulations aimed at restricting China's access to cutting-edge computer chips and promoting investments within the United States for national security purposes. During a recent confirmation hearing, Howard Lutnick, who has been nominated for the position of Commerce Secretary, echoed concerns regarding intellectual property theft and hinted at the possibility of further actions to protect American AI firms. He remarked, 'What this showed is that our export controls, not backed by tariffs, are like a whack-a-mole model. ' OpenAI has publicly stated that companies from China and other regions are consistently attempting to extract knowledge from the models developed by leading US AI firms. They emphasized the critical importance of collaborating closely with the US government to ensure the protection of their most advanced models as they move forward. Naomi Haefner, an assistant professor of technology management at the University of St. Gallen in Switzerland, raised questions about the validity of DeepSeek's claims regarding the low cost of developing its application. She stated, 'It is unclear whether DeepSeek really trained its models from scratch. ' OpenAI has suggested that DeepSeek may have misappropriated substantial amounts of data from their systems. If this assertion holds true, it would cast doubt on DeepSeek's claims of achieving cost-effective model training. Until a thorough examination of DeepSeek's training methods is conducted, it remains uncertain whether such economical training is genuinely feasible. Crystal van Oosterom, an AI Venture Partner at OpenOcean, concurred that DeepSeek has evidently built upon publicly available research from prominent American and European institutions and companies. However, the ethical implications of 'building on' the work of others in the AI sector remain ambiguous, particularly given the frequent accusations of violating intellectual property rights. US officials are also contemplating the national security ramifications of DeepSeek's emergence, as indicated by White House press secretary Karoline Leavitt. She mentioned that the National Security Council is actively investigating the potential national security implications associated with DeepSeek's activities. Leavitt reiterated President Donald Trump's earlier remarks, suggesting that DeepSeek should serve as a wake-up call for the American technology industry. This announcement follows reports that the US Navy has prohibited its personnel from using DeepSeek's applications due to potential security and ethical concerns. According to CNBC, the Navy has issued a warning to its staff, advising them against using the DeepSeek app due to the associated risks. The Navy has not yet responded to inquiries from BBC News regarding this matter. Experts in data security have cautioned users to exercise caution when using DeepSeek, as the application collects significant amounts of personal data and stores it on servers located in China. Meanwhile, DeepSeek has claimed that it is currently facing cyber attacks. On Monday, the company announced that it would temporarily restrict new registrations due to 'large-scale malicious attacks' targeting its software. A notice displayed on the company's website indicates that registration may be delayed as a result of these attacks.