OpenAI and Microsoft and the impact on news publishers

Recently released court documents from the lawsuit of The New York Times against OpenAI and Microsoft reveal that these companies were aware of the risks of their AI training on web data early on but continued their practices nonetheless. Internal reports indicate that data collection is harmful to the open web and news publishers, with a shift in user behavior where chatbots replace traditional searches. This results in a lower click-through rate to original sources.

Brent Hecht, Director of Applied Science at Microsoft, refers to the data collection for AI models as a "stunning theft of unprecedented proportions" and points to the potential self-harm that this development brings. Microsoft CEO Satya Nadella has testified under oath that AI chat analyses have essentially already replaced search engines, undermining the need for people to seek out the original source.

At OpenAI, there are similarly concerning internally made remarks. The head of ChatGPT calls the impact of AI products on news publishers an "existential threat" and states that chatbots primarily function as replacements. This raises questions about the fair-use defense that emphasizes the transformative value of AI content versus the harm it causes to the news industry.

Additionally, OpenAI acknowledged that their GPT models are capable of literally reproducing existing content. This raises concerns about the treatment of paywall content, as it appears OpenAI has made no efforts to identify or remove this content from their training data. The broader implications of this situation affect not only large tech companies and news publishers but also the entire ecosystem of AI services. The ruling of the judge in this case will be crucial for determining the future of data collection by AI and potential licensing requirements.

Read the full article from AI Insider.