Black Inc. Seeks AI Partnerships, Raising Concerns Among Authors
Melbourne-based publisher Black Inc Books has initiated a controversial move, seeking partnerships with unnamed artificial intelligence (AI) companies. The publisher is requesting its authors’ permission to utilize their written works for training AI systems.
According to reports, writers are being asked to grant Black Inc. certain copyright rights to facilitate the development of machine learning and AI. This permission would allow the use of authors’ works for the training, testing, validation, and eventual deployment of AI systems.
A Black Inc. spokesperson indicated that this contract is offered on an opt-in basis. The company also stated that it would negotiate with “reputable” AI companies.
However, this proposal has drawn criticism from various fronts including authors, literary agents, and the Australian Society of Authors. Novelist Laura Jean McKay articulated her concern, stating, “I feel like we’re being asked to sign our own death warrant.”
The Growing Trend of AI-Publisher Alliances and Author Apprehensions
Such partnerships between AI companies and publishers could potentially become more commonplace. Non-fiction authors may have particular reason for concern. While there may remain a market for authentic fiction from human writers, the situation still presents challenges.
Large language models utilize probability-driven algorithmic language systems. Improvements in these models continue at an exponential rate, which has considerable ramifications for writers.
Publishers, AI, and the Licensing Dilemma
Publishers are seeking to partner with AI companies employing a similar approach to the music industry and other content producers when confronted with transformative technologies like Spotify and YouTube. These platforms were initially viewed as threats to established copyright interests.
Licensing agreements provided a solution, permitting usage in exchange for recurring fees. Although disputes around fairness in licensing fees sometimes arise, as seen in the controversy surrounding Spotify’s royalties for musicians, this licensing model has largely allowed content industries and new technologies to coexist.
Ideally, a licensing solution should also work for authors, publishers, and AI companies, potentially generating revenue streams. However, the actual income for authors could be minimal, and the legal basis for securing it might be tenuous.
Copyright Law’s Limitations in the Age of AI
Authors and publishers rely on copyright law to protect their interests. However, copyright law focuses on copying, not on the development of capabilities in probability-driven language outputs. The needs of AI systems:
Data requirements for AI systems are substantial. Training AI necessitates massive amounts of data to reach the standard required to imitate or replace human authors. In the initial phase of the AI revolution, AI companies appear to have simply collected copyright-protected content.
Now, publishers and organizations that collect fees on behalf of copyright holders are fighting back. Yet, copyright battles between AI companies and publishers are far from resolved.
A group of Canadian news organizations has filed a lawsuit against the maker of ChatGPT for copyright infringement. In the United States, The New York Times v. OpenAI case is ongoing.
The Need for Global AI Regulations
In the realm of AI copyright infringement cases, the legal outcomes remain uncertain since a superior court has yet to issue a definitive judgment.
While governments, including Australia’s, are considering the issue, no international consensus on the rules exists. The European Union is leading the race to establish regulations for AI which is why US vice-president J.D. Vance was so eager to shut down the possibility of EU regulation expanding to other jurisdictions. The bigger issue remains data. AI systems need human writing to learn from.
The Questionable Benefits for Authors
AI companies thus find it logical to collaborate with publishers. This approach avoids copyright disputes and provides access to data. But does partnering with AI make sense for writers?
Copyright protects expression. Copyright law governs who can copy that expression and under what conditions. Copying is involved in the development of AI systems. However, it is unclear whether actual copying takes place after the AI system has learned from the data.
Black Inc.’s Contract Offer and Concerns
Black Inc. is essentially seeking a contract modification with its authors. According to The Guardian, the publisher will divide net receipts from AI deals with the author 50/50.
Kate Nash, head of marketing and publicity at Black Inc., stated, “We believe authors should be credited and compensated appropriately and that safeguards are necessary to protect ownership rights in response to increasing industrial automation.”
However, while publishers frequently discuss licensing, they are less clear about why AI companies should continue paying them.
AI technology focuses on the development of capabilities; it does not operate on the basis of ongoing copying of protected material. Therefore, it is not like music on YouTube, Spotify, or other creative content on platforms such as Google. Once the AI has learned how to write, it has gained that skill.
AI can be influenced to produce output that includes copyright-protected content. However, prompting the AI to reproduce protected content typically is not its standard operational procedure, so it not a strong basis for a copyright infringement claim.
Ultimately, the issue, as McKay suggests, is that AI systems could potentially outcompete some authors, potentially creating problems for AI companies that diminishes the available data and devastates authors who lose their livelihoods.
This industrial problem requires government intervention to establish a system that balances innovation and rights.
