News media publishers, as a group, made what turned out to be a costly error in the early days of the internet. Excited at the prospect of gaining uncounted new readers beyond their home markets they made their reporting openly available online for free. What they did not, and perhaps could not anticipate in those early days, was the monopolistic potency of network effects that would shape the online ecosystem. The result was to concede publishers’ direct connection with readers to search engines and social media platforms and the rupture of their relationship with advertisers.
Ever since, news publishers have been fighting to regain lost ground, with limited success. Those with readership large enough to make the math work increasingly have erected paywalls. But shifting costs long borne by advertisers to readers has its limits and even most paywalled news sites remain dependent on non-paying readers referred by Google, Facebook and other intermediaries.
More recently, publishers have sought relief from regulators and policymakers, pushing for legislative changes to require online platforms to secure and pay for licenses from publishers to display, distribute or otherwise repurpose publishers’ content. Those efforts have found success in some territories, notably Australia, the European Union and Canada. But the licensing systems that have emerged, to the extent they’ve been effective, have generally benefitted the largest publishers with the heft to negotiate directly with platforms, rather than the small and local publishers who arguably need the help more.
Paywalls and payments from platforms, moreover, may now be having the paradoxical effect of contributing to the general erosion in the value of professionally produced news reporting on social media platforms. Publishers across the board have recently reported declines in search and social referral traffic as users turn to other sources for information and advertisers drift away to less news-focused platforms like TikTok and Instagram.
Having effectively lost the battle of search and social media, news publishers are now facing a new threat in generative artificial intelligence. This time, though, publishers are alert to the danger and have taken steps to avoid repeating past mistakes.
In the 12 months since OpenAI’s ChatGPT appeared last November, many leading publishers have moved to erect barricades against the free use of their content by AI platforms. A recent survey of 1,149 news publishers by data journalism site Palewhite found that 551, or 48% of the total, including the New York Times, the Washington Post, the Wall Street Journal, Reuters and Bloomberg, have already moved to block OpenAI from scanning their websites to train the next iteration of its generative AI model. More than a third (410) have also moved to block Common Crawl, the non-profit dataset scraped from the web and used by OpenAI and other gen-AI developers.
Barry Diller-owned IAC, which owns The Daily Beast, has sought to organize a joint effort among leading news publishers to bring one or more more copyright infringement lawsuits against AI developers over the use of publishers’ content to train their models. While that effort apparently has sputtered, the New York Times has threatened unilateral action against OpenAI after negotiations between the two collapsed.
The News|Media Alliance, which represents more than 2,000 print and digital publishers, has organized Capitol Hill visits by its members to press congress for legal protections against unauthorized scaping of their content, among other issues.
News publishers also have allies in the fight, or at least other enemies of their enemies, in other media sectors. Book authors and publishers, visual artists, music publishers and other rights owners have filed lawsuits against a range of AI developers and have also been active in pressing policymakers for protections against unauthorized use of their content.
Like generals fighting the last war, however, news publishers may be guarding against the wrong threat. Whatever else their impact, search and social media were simply new models for performing familiar functions: the copying, distributing, performing and displaying expressive works. Generative AI is an altogether different phenomenon. It does not, in and of itself, exploit the expressive content of the works it makes use of, either in training the model or in the output it produces. Even the apparent, or frankly ineluctable similarity of its output to pre-existing works is not a function of copying per se but of the operation of incomprehensively complex probabilistic functions.
Locking down news content then, even if it compels AI companies to pay for access, can at best produce only stalemate. Meanwhile, generative AI technology and models will continue to evolve, with or without professional journalism. And it will change the information environment in fundamental and lasting ways, more than search and social media ever could.
The closest analogy might be climate change. Even if we were to stop spewing carbon into the atmosphere today, humans (along with every other species) will have to cope with its effects for millennia, if we last that long. So, too, with generative AI. Even if today’s media organizations manage to gain a measure of control over the use of their content by AI systems, they still must figure out how to cope with its effects if they’re to survive.