- Recent AI search advancements by OpenAI and other tech leaders are prompting news organizations to reassess potential partnerships with AI firms.
- Negotiations are shifting from broad data usage for language models to more specific applications where news publishers might gain leverage.
- Language models, trained on extensive online datasets, require curated, real-time news data to provide accurate information through Retrieval Augmented Generation (RAG).
- OpenAI’s SearchGPT and Microsoft’s Bing generative search product are evolving how tech and media firms collaborate, with features like inline attribution and content management tools for publishers.
- Uncertainty exists over whether new AI search engines will match the revenue generated by traditional search engines like Google.
- OpenAI is exploring revenue-sharing models, while current agreements with news publishers are based on licensing fees; startups like Tollbit are working on revenue-sharing marketplaces.
- Legal debates over copyright and public data usage are ongoing, with significant cases such as The New York Times’ lawsuit against OpenAI and Microsoft potentially shaping future regulations.
- Some AI companies, including Anthropic and Meta, are cautiously navigating media partnerships, while Perplexity faces resistance from media firms over content use.
Main AI News:
Recent advancements in AI search technologies by OpenAI and other industry leaders are compelling news organizations to reevaluate their strategies and potential partnerships with AI firms that require news content for real-time event queries.
The broader context: While initial discussions between tech companies and media organizations centered on supplying data for training expansive language models, current negotiations are shifting towards more specific applications. This change could offer news publishers increased bargaining power.
Mechanics of AI search: Language models are trained using extensive datasets from diverse online sources, but to provide accurate, up-to-date information, these models need access to curated, real-time news data. This process, known as Retrieval Augmented Generation (RAG), enhances model accuracy and mitigates the risk of incorrect answers, although it cannot entirely eliminate them.
News dynamics: The recent launches of OpenAI’s SearchGPT and Microsoft’s Bing generative search product have highlighted the evolving nature of partnerships between major tech companies and news publishers. OpenAI is currently collaborating with news organizations such as The Atlantic and News Corp to refine SearchGPT. The platform includes features like clear, inline attribution and quick access to additional sources, while publishers have tools to manage their content’s visibility.
Challenges ahead: There is uncertainty about whether new generative AI-powered search engines will generate revenue comparable to traditional search engines like Google. Historically, Google’s model of directing traffic to publishers via search links has been effective. Although OpenAI is exploring revenue-sharing models for its GPT store, current agreements with news publishers involve licensing fees. Meanwhile, startups like Tollbit are developing marketplaces for shared revenue, though these require broad engagement to succeed.
Legal landscape: AI companies claim legality in training models on publicly available content, yet many publishers contend that their content is copyrighted. The ongoing lawsuit by The New York Times against OpenAI and Microsoft may clarify these legal issues, though a resolution could take years. For now, both news organizations and AI companies are navigating RAG agreements while avoiding new copyright disputes. Notably, OpenAI allows news sites to appear in SearchGPT results even if they opt out of generative AI training.
Future outlook: While OpenAI actively seeks news partnerships, other AI firms are more cautious. Anthropic has not confirmed any agreements with publishers, and Axios reports no known deals with the company. Meta is deliberating its approach, with internal debates over the value of media partnerships. Despite past fluctuations in media relationships, Meta may need to collaborate with news providers for accurate AI functionalities. Perplexity, another player in the field, faces resistance from media firms over its use of content and has encountered legal threats from companies like Forbes and Condé Nast.
Conclusion:
The emergence of advanced AI search technologies is significantly altering the media landscape, with new models and partnerships reshaping the dynamics between tech firms and news publishers. As these technologies integrate more deeply into search engines, media companies may find new opportunities to negotiate better terms and revenue structures. However, the legal and financial uncertainties surrounding AI-generated content and copyright issues will continue to influence how these partnerships evolve, requiring both industries to adapt to rapidly changing conditions.
