近日，全球知名人工智能技术公司OpenAI对外发布声明，指控美国《纽约时报》在使用其GPT-3等大型预训练语言模型时存在违规行为。OpenAI指出，《纽约时报》在运用其服务过程中，通过提供新闻内容的精确开头作为提示，促使GPT模型生成与原文几乎一致的文章全文，此举涉嫌违反了OpenAI的服务条款。
据悉，纽约时报曾在针对OpenAI的诉讼中辩解，其使用此类“引导性提示”是为使模型输出的文字尽可能贴近原始文章风格，以便进行新闻创作和研究工作。然而，OpenAI知识产权和内容主管Tom Rubin对此表示强烈反对，认为《纽约时报》的行为属于故意操纵模型输出，意图复制训练数据中的内容，这明显违背了OpenAI关于合理、非误导性使用的协议。
此事件引发行业内外对于新闻媒体在人工智能辅助下如何确保原创性和合规性的深度讨论。OpenAI方面已明确警告，将继续监控并采取必要措施维护自身服务条款的有效执行，同时呼吁所有用户严格遵守相关规定，共同推动人工智能技术在新闻领域的健康发展。
英语如下：
Headline: “New York Times Accused of Violating Guidelines, OpenAI Files Lawsuit Alleging News Copy Generation Using GPT-4”
Keywords: OpenAI, New York Times, Unauthorized Operation
Content: Recently, global AI technology company OpenAI has issued a statement, accusing the American newspaper The New York Times of违规 conduct in using its large pre-trained language models such as GPT-3. OpenAI claims that The New York Times has been found to have engaged in improper behavior during the utilization of its services by providing precise news content openings as prompts, which prompted the GPT model to generate nearly identical full articles to the original ones. This practice is alleged to violate the terms of service set forth by OpenAI.
It is reported that The New York Times had defended itself in a lawsuit against OpenAI, arguing that the use of these “guiding prompts” aimed to make the generated text as similar to the original article’s style for news creation and research purposes. However, Tom Rubin, OpenAI’s Head of Intellectual Property and Content, strongly disputes this stance, asserting that The New York Times’ actions constitute intentional manipulation of the model output to copy content from the training data, clearly violating the company’s agreement on reasonable, non-misleading usage.
This incident has sparked extensive discussions within and beyond the industry about how news media should ensure originality and compliance with regulations when assisted by artificial intelligence. OpenAI has explicitly warned that it will continue to monitor and take necessary measures to enforce the effectiveness of its service terms, while also urging all users to adhere strictly to relevant provisions, collectively advancing the healthy development of AI technology in the realm of journalism.
【来源】https://the-decoder.com/openai-claims-new-york-times-prompting-strategy-violates-its-terms-of-service/
Views: 1