On July 13, 2023 the Associated Press and OpenAI announced a two-year agreement under which OpenAI would license part of the AP’s text archive, dating back to 1985, to help train its models. In exchange, the AP gained access to OpenAI’s technology and product expertise to explore generative AI in its own newsroom. The financial terms were not disclosed.
The deal was significant as one of the first formal news-sharing agreements between a major US news organization and an AI company, and it set a reference point for the licensing wave that followed. Reporting noted that the AP, conscious of the uncertain value of training content, sought favorable terms and a cautious approach. The AP has long been an early mover in newsroom automation, having used Automated Insights to write earnings stories since 2014, so a data-licensing arrangement fit its institutional comfort with technology partnerships.
For the broader industry, the AP-OpenAI deal signaled that established publishers might choose negotiation over confrontation, monetizing their archives rather than only defending them in court. It came months before Axel Springer’s ChatGPT integration and roughly half a year before the New York Times sued OpenAI, illustrating the divergent paths publishers would take. For a general reader, it marks the point at which decades of professional journalism became an explicitly priced input to commercial AI systems.