OpenAI and Associated Press (AP) announce partnership to train AI on news articles

In a landmark deal announced today, generative AI pioneer OpenAI is partnering with the Associated Press (AP), one of the world’s oldest and most well-read newswire services, established in 1846. 

The partnership is the first of its kind between a major AI vendor and media outlet, and is in some ways the meeting of two giants of their respective industries. It will see OpenAI licensing text content from the AP archives that will be used for training large language models (LLMs). In exchange, the AP will make  use of OpenAI’s expertise and technology — though the media company clearly emphasized in a release that it is not using generative AI to help write actual news stories.

>>Follow VentureBeat’s ongoing generative AI coverage<<

The deal is important for both vendors as a bridge that will help to bring the media icon into the generative AI era on one side, and on the other as a means to inform OpenAI’s large language models (LLMs) with the depth of human-authored news intelligence and expertise from the AP. The news also comes mere hours after the U.S. Federal Trade Commission, the agency in charge of overseeing industry for consumer harms, was reported to be investigating OpenAI over a data breach that exposed customer payment information and chat histories, as well as inaccuracies.

While the partnership offers tremendous opportunity for both firms, there is no specific dollar figure attached to it, as financial terms of the deal are not being publicly disclosed. While OpenAI’s ChatGPT LLM has achieved nearly 200 million monthly users in recent months, the AP counts a readership of four billion people daily through its many deals syndicating its content to thousands of local and national news outlets in the U.S.

OpenAI commits to ‘supporting vital work of journalism’

The partnership between OpenAI and the AP comes at a particularly critical time in the development of the nascent generative AI industry, amid ongoing pressure facing the news business from emerging technology.

OpenAI, as well as other generative AI vendors, have been criticized in recent weeks and months for scraping publicly available data and using it to train their models without the express informed consent of the data creators. Currently OpenAI is facing multiple legal challenges over data scraping, where plaintiffs allege that the AI vendor used content without authorization or the legal right to do so. Just last week, comedian Sarah Silverman joined the chorus of complaints with a lawsuit of her own, alleging copyright infringement.

Protecting copyright and valuing the news industry are critical aspects of the partnership between OpenAI and the AP.

“We are pleased that OpenAI recognizes that fact-based, nonpartisan news content is essential to this evolving technology, and that they respect the value of our intellectual property,” Kristin Heitmann, AP senior vice president and chief revenue officer, said in a statement. “AP firmly supports a framework that will ensure intellectual property is protected and content creators are fairly compensated for their work.”

Heitmann added that from her perspective it’s critical that all news organizations are part of the conversation when it comes to generative AI, to help ensure that the journalism industry can benefit from the technology.

For its part, OpenAI is showing its ability to collaborate and support industries that it is being accused of disrupting, claiming that the partnership is indicative of its support for journalists.

“OpenAI is committed to supporting the vital work of journalism, and we’re eager to learn from the Associated Press as they delve into how our AI models can have a positive impact on the news industry,” Brad Lightcap, chief operating officer at OpenAI, wrote in a statement.

AI isn’t new for the AP

While the AP is not currently using generative AI to write stories, it is no stranger to the world of artificial intelligence.

The AP claims that for the past decade it has used AI in different capacities to automate tedious tasks, helping its journalists to be more productive. For example, the AP has been using non-generative forms of AI since 2014 to help automate the process of composing corporate earnings reports.

AP has also used AI for voice and video transcription.

Search is yet another area where AP is using AI. On June 1, AP launched an AI-powered search function, using technology from software vendor MerlinOne, to help users more easily find images and videos.

Neither OpenAI nor the AP specified if the new partnership would intersect with or remain independent of these other efforts, but clearly, the AP sees value in AI — and now we know that for the leading consumer AI service provider, the feeling is mutual.

Source