\n<\/p>\n
OpenAI\u2019s legal<\/a> battle<\/a> with The New York Times over data to train its AI models might still be brewing. But OpenAI\u2019s forging ahead on deals with other publishers, including some of France\u2019s and Spain\u2019s largest news publishers.<\/p>\n OpenAI on Wednesday announced<\/a> that it signed contracts with Le Monde and Prisa Media to bring French and Spanish news content to OpenAI\u2019s ChatGPT<\/a> chatbot. In a blog post, OpenAI said that the partnership will put the organizations\u2019 current events coverage \u2014 from brands including El Pa\u00eds, Cinco D\u00edas, As and El Huffpost \u2014 in front of ChatGPT users where it makes sense, as well as contribute to OpenAI\u2019s ever-expanding volume of training data.<\/p>\n OpenAI writes:<\/p>\n Over the coming months, ChatGPT users will be able to interact with relevant news content from these publishers through select summaries with attribution and enhanced links to the original articles, giving users the ability to access additional information or related articles from their news sites \u2026 We are continually making improvements to ChatGPT and are supporting the essential role of the news industry in delivering real-time, authoritative information to users.<\/p>\n<\/blockquote>\n So, OpenAI\u2019s revealed licensing deals with a handful of content providers at this point. Now felt like a good opportunity to take stock:<\/p>\n How much is OpenAI paying each? Well, it\u2019s not saying \u2014 at least not publicly. But we can estimate.<\/p>\n The Information reported<\/a> in January that OpenAI was offering publishers between $1 million and $5 million a year to access archives to train its GenAI models. That doesn\u2019t tell us much about the Shutterstock partnership. But on the article licensing front \u2014 assuming The Information\u2019s reporting is accurate and those figures haven\u2019t changed since then \u2014 OpenAI\u2019s shelling out between $4 million and $20 million a year for news.<\/p>\n That might be pennies to OpenAI, whose warchest sits at over $11 billion and whose annualized revenue recently topped $2 billion (per<\/a> Financial Times). But as Hunter Walk, a partner at Homebrew and the co-founder of Screendoor, recently mused, it\u2019s substantial enough to potentially edge out AI rivals also pursuing licensing agreements.<\/p>\n Walk writes<\/a> on his blog:<\/p>\n [I]f experimentation is gated by nine figures worth of licensing deals, we are doing a disservice to innovation \u2026 The checks being cut to \u2018owners\u2019 of training data are creating a huge barrier to entry for challengers. If Google, OpenAI, and other large tech companies can establish a high enough cost, they implicitly prevent future competition.<\/p>\n<\/blockquote>\n Now, whether there\u2019s a barrier to entry today is debatable. Many \u2014 if not most \u2014 AI vendors have chosen to risk the wrath of IP holders, opting not to license the data on which they\u2019re training AI models. There\u2019s evidence that art-generating platform Midjourney, for example, is training<\/a> on Disney movie stills \u2014 and Midjourney has no deal with Disney.<\/p>\n The tougher question to wrestle with is: should licensing simply be the cost of doing business and experimentation in the AI space?<\/p>\n Walk would argue not. He advocates for a regulator-imposed \u201csafe harbor\u201d that\u2019d protect any AI vendor \u2014 as well as small-time startups and researchers \u2014 from legal liability so long as they abide by certain transparency and ethical standards.<\/p>\n Interestingly, the U.K. recently tried<\/a> to codify something along those lines, exempting the use of text and data mining for AI training from copyright considerations so long as it\u2019s for research purposes. But those efforts ended up falling through.<\/p>\n Me, I\u2019m not sure I\u2019d go so far as Walk in his \u201csafe harbor\u201d proposal considering the impact AI threatens to have on an already-destabilized news industry. A recent model from The Atlantic found<\/a>\u00a0that, if a search engine like Google were to integrate AI into search, it\u2019d answer a user\u2019s query 75% of the time without requiring a click-through to its website.<\/p>\n But perhaps there is<\/em> room for carve-outs.<\/p>\n Publishers should be paid \u2014 and paid fairly. Is there not an outcome, though, in which they\u2019re paid and challengers to AI incumbents \u2014 as well as academics \u2014 get access to the same data as those<\/em> incumbents? I should think so. Grants are one way. Larger VC checks are another.<\/p>\n I can\u2019t say I have the solution, particularly given that the courts have yet to decide whether \u2014 and to what extent \u2014 fair use shields AI vendors from copyright claims. But it\u2019s vital we tease these things out. Otherwise, the industry could well end up in a situation where academic \u2018brain drain\u2019 continues unabated and only a few powerful companies have access to vast pools of valuable training sets.<\/p>\n<\/p><\/div>\n\n
\n
\n