_(44538272181).jpg)
For those who peruse Twitter/X, Jack Dorsey’s recent call to “delete all IP law” might have melded into the mass of other dull, odd statements made on the platform. However, his statement, and further clarifications, touch on an important crisis unfolding in copyright law and reflect a common nonconformist attitude among players in emerging industries.
The electric scooter and bicycle-sharing companies famously did not seek approval before introducing their products and services into various American localities. The CEO of Bird, in 2018, even publicly professed that in the places “where there’s no laws [sic], that’s where we go in.” While the White House claims that Tren de Aragua, the Venezuelan gang, has launched a predatory incursion against the United States, the overnight invasion of American cities by electric scooters feels closer to fitting the bill.
The development of generative models such as ChatGPT is a technological marvel; the ability to efficiently generate scores of articles, images, and other content will revolutionize industry and fundamentally change the way we think about human creativity. However, it is also true that, like the e-scooter companies, many GenAI leaders have embraced the “act now, apologize later” approach to doing business. While job displacement, deepfakes, and great power competition have dominated the conversation about AI, the threat of AI to copyright remains a lesser-known issue.
Authors, artists, and other creators, however, are very aware of the indelible and extensive breaches of copyright law which have occurred. Protests recently broke out in London and Dublin against Meta in light of court revelations wherein plaintiffs discovered that Mark Zuckerberg, to train Meta’s Llama 3 model, authorized a raid on LibGen — a Russian website with a treasure trove of millions of pirated, copyrighted works. With the widespread attacks on vessels transiting the Red Sea, it would appear that piracy is back in style.
“If the PRC’s developers have unfettered access to data and American companies are left without fair use access, the race for AI is effectively over. America loses, as does the success of democratic AI.”— OpenAI’s Public Comments on Trump’s AI Action Plan
With lawmakers comparing the release of DeepSeek to Sputnik, and with the Vice President urging Europeans not to cripple AI through over-regulation, the call for slashing copyright as a tool for defeating the Chinese is sure to strike a chord with many in Washington.
Should we buy into the demand by AI companies to ratify the nonconsensual use of copyrighted works? Should we go even further and embrace the anti-IP vision of the brainiacs from Silicon Valley like Jack Dorsey, Elon Musk, and Marc Andreessen?
The tension between copyright and emerging technologies is not new. In fact, for all its fanfare and praise, AI as a technology is, in large part, not even new. As an exercise, go ask an expert in signal processing, optimization, or computational linguistics about AI. The common refrain one hears is: “I was doing that before it was cool” or “I was doing that before it was called AI or machine learning.”
Back in the 1900s, a battle had emerged between composers and the player piano: a self-playing piano that could perform famous musical works using strange, perforated sheets. In the 1970s, another battle commenced between publishers and the photocopier as the product saw widespread adoption in offices. In both cases, users of the new inventions heedlessly reproduced copyrighted works while the companies that purchased or developed the technology sat idly by, refusing to compensate the rights-holders.
Congress responded differently to each case. In the wake of White-Smith v. Apollo, lawmakers created a compulsory licensing system whereby any person could make a phonorecord of a copyrighted composition, with some exceptions, by paying a statutory royalty and filing notice with the rights-holder and Copyright Office. Part of the impetus for the policy was the expanding monopoly of the player piano company, Aeolian, over performing rights.
The Copyright Act of 1976 did not emulate the Copyright Act of 1909; instead, Congress told the parties involved in the photocopier dispute to develop a voluntary, simple, cost-effective solution to the problem. Publishers responded to the suggestion by creating the Copyright Clearance Center — a collective management organization.
Collective licensing is now a well-known concept. A mass of rights-holders transfer management of one or more rights over to a central entity, known as a collective management organization (CMO), that leases those rights to interested parties. The license provides legal security to companies wishing to exploit the protected works and remunerates the rights-holders for their contribution.
Can collective licensing solve the AI problem? Some CMOs have already started rolling out licenses that permit AI re-use rights. However, there are many issues with this approach. To address an earlier point, it is true that American companies face obstacles that global competitors, like Chinese firms, do not. At the same time, it is also true that the United States has the power to write the rules of global trade and commerce — a power we lose in an increasingly hostile, fragmented, and protectionist world. Would you prefer multilateral cooperation in solving cross-border issues or the Dorsey-sponsored rat race?
There are other problems, too. While creators embrace the collective licensing approach to AI, many also advocate remuneration for prior unauthorized use — a natural but naive demand. Other demands are sometimes tossed into the mix such as a federal right of publicity and compulsory disclosure of datasets and AI-generated content. Remuneration for past use will likely never occur. Besides the dataset concern, the other demands are only tangentially related to the conversation. Perhaps we can bypass this issue by conceding that the pro-copyright lobbyists and activists, in order to form consensus in Washington, will have to bite the bullet on many of their demands.
Additionally, as Dorsey correctly points out, there are long-standing problems with transparency on how rights-holders are compensated. The Copyright Office is currently soliciting comments on the royalty distribution practices of performing rights organizations following a request from Congress. Jonathan Band, counsel to the Library Copyright Alliance and adjunct professor at Georgetown, has published an article on the long history of unscrupulous behavior by CMOs around the globe.
In spite of their abuse of market dominance, CMOs are the best voluntary, private solution that upholds copyright law. In fact, while seemingly contradictory, CMOs work best — and arguably only work — when they are few in number. What is the point of collective licensing if there are numerous collective licensing agencies? As experts at the Max Planck Institute for Innovation and Competition point out, monopolistic dominance by a CMO may very well represent efficient market behavior.
It is ironic that the hero to save us from tech monopolies is a monopolistic agency that upholds monopoly rights. Thomas Jefferson is surely rolling in his grave. In our strange world, up is down, left is right, but Jack Dorsey is still wrong.