Confessions of a Content Licensing Lawyer
(Or Fasten Your Seat Belts)
Frank Bilotto, Esq., Creative Licensing International’s Business and Legal Affairs member, shares some facts and debunks some myths from his 25 years of licensing content for both licensors and licensees. He also addresses challenges that content creators will have with distribution and royalties in the age of generative AI.
I’ve been licensing content for 25 years. Initially, I represented licensees, where my role was to pay as little as possible for the widest distribution of content. Ten years ago, I switched teams. Since then, I’ve represented publishers.
While my role as a lawyer includes protecting my clients against any undue risk related to the dissemination of their content, my primary responsibilities are to ensure that my clients are properly paid for the usage of their content and to restrict the distribution of their content only to places where they desire to have their content distributed and only by organizations that they wish to distribute their content.
If you think this sounds easy, keep in mind that when I worked for the largest aggregator and redistributor of content in the world, Dialog (before it was acquired by Thomson Reuters), we had complex algorithms to determine what to pay publishers. Does anybody remember Dial Units? No one, not even Dialog employees, knew how they were calculated. Hence, we never lost an audit challenge from a publisher.
Additionally, we had dozens of redistributors and agents that sent the content that I licensed to customers of which I was completely unaware. More importantly, the publisher was also completely unaware of where their content was being distributed.
Today, my job is to negotiate agreements and draft language that prevents unwanted distribution and underpayment for usage of content. For 50 years in what was once known as the redistribution of electronic content and now digital content, there was always a commonly understood practice that usage of content would be by individuals that consume the content for research or pleasure.
Users searched for and accessed content one article at a time. Then if an individual desired to repurpose the content in any way, it would only be within the rights granted by the redistribution agreement and in exchange for appropriate fees to be paid. While there were some exceptions, I could always be confident that my contracts not only protected my clients, but also generated the most income they could possibly earn from licensing.
Generative AI Changes the Game
However, about a year ago, I was caught completely unprepared for the disruptive force of ChatGPT and other machine learning technologies. As such, my job became exponentially more difficult. The basic premise of how AI works has completely changed the game. AI doesn’t consume one article at a time but ingests the entire archive of a publication into a machine learning engine for future use in creating answers to questions presented by customers.
I’m the first to admit that my existing content licenses did nothing to prevent this type of content consumption. In my agreements, if a customer accesses the entire archive of a publication, that customer would pay for the content consumed.
But in the case of AI, not only is the content consumed, it is also stored, essentially becoming the property of the machine for multiple future uses, which makes it difficult, if not impossible, to account for all the uses of the content going forward. Even if the publisher is paid for the one-time download of the entire archive, there currently is no accounting for the use of that content in derivative works by the AI. The publisher has no knowledge or control of the distribution of its content. More troubling is that a publisher would be unlikely to recognize its own content in a derivative work created by generative AI. All of this makes it impossible for the publisher to be fairly paid or limit distribution.
So, what can a publisher and a content licensing lawyer like me do? First and foremost, anyone who tells you they are an AI content licensing expert is either naive or lying. Everything is too new, and not much has been done legally to address the problem.
All legal action and conversation is focused on copyright violation, which doesn’t specifically address licensing. Making publishers’ problems with AI more difficult, Google announced on October 12th that it will defend users of generative artificial-intelligence systems in its Google Cloud and Workspace platforms if they are accused of intellectual property violations, joining Microsoft, Adobe, and other companies that have made similar pledges.
Now the big guns of the content distribution world have taken dead aim at publishers. Clearly, Google doesn’t remember the business model they implemented to encourage publishers to allow their content to be indexed by Google. Google was a search engine that benefitted publishers by redirecting traffic back to the publisher’s website, so that the publisher could monetize their content with advertising and sponsorships. The model made money for both parties. Google could monetize the search process and the publisher could monetize the content.
Google Ushers in New World of Content Distribution
But that all changed when Google began to use early AI technology. For at least 12 years, Google has been providing answers to search queries on Google’s home page, using a publisher’s content to provide the answer without ever redirecting the traffic to the publisher’s web site. This was just the beginning of Google changing the world of content distribution and not properly compensating the publisher for use of content. Google’s initial motto of “Don’t Be Evil,” subsequently replaced in 2015 by “Do the Right Thing,” evidently didn’t include not cutting off revenue streams from publishers.
“Fair use” is a term that is commonly used for what Google does. It began with snippets and evolved into posting answers to queries by directly scraping content, like football scores and flight information. If left to the standards established by the courts, fair use would never be specifically defined to my satisfaction. As a lawyer representing publishers, fair use is specifically defined in the agreements we negotiate with licensees.
AI has made traditional fair use abuses seem tame. While I was initially unprepared for AI, I took immediate action with all of my licensees with attempts to renegotiate terms of the agreements to include specific limitations on AI usage. Candidly, most delayed the negotiations or ignored the requests completely. My only recourse with those licensees is to address AI issues at the time of renewal.
In any case, in every renegotiation and new license agreement, I have already prepared specific language of how content can be utilized by AI in accordance with the desires of my client publishers. And I’ve done it within the guidelines of what I consider my primary responsibilities — to control the distribution of and maximize the revenue for the content of the publisher.
Cannibalization Is Real — and Lethal
I remember 10 years ago, when publishers were fearful of the cannibalization of their content. I was the loudest voice in the room trying to convince publishers that cannibalization was a myth. I was proven right by the publishers that followed my advice, as they watched their revenues increase. However, with AI, cannibalization is real and it’s lethal. I can confidently predict that in the coming years many publishers will face bankruptcy because they did not properly protect their content against exploitation by AI technologies.
Look, I’m no AI expert. I still have a great deal to learn. I don’t even know what I don’t know yet. Neither does anyone else.
AI is an infant. No one knows what AI will be when it grows up, but it will grow up to become an integral part of content creation. I only know that when AI grows up, whatever it becomes, I’ll still be controlling the distribution of my clients’ content and maximizing their revenues. But it’s not going to be easy. To quote Bette Davis in the movie All About Eve, “Fasten your seat belts, it’s going to be a bumpy night.”
Coda to Confessions
By Frank Bilotto
I was surprised and pleased to receive feedback from 4 continents on my article, Confessions of Content Licensing Lawyer. Some challenged my assertions, some thanked me, and some asked me specifically for the restrictive language I use.
Just a few comments.
For those that asked for specific language, while I didn’t send copies of contracts to those that requested it, I am happy to share my thinking on this issue.. Below is the verbatim answer that I sent to a European licensing lawyer.
Before I go any further, I must tell you that while I believe the language in my agreements sufficiently protects content from the download of entire archives for machine learning, I am not confident that aggregators have sufficient means to prevent those type of transactions. As such, I have included liquidated damages for such occurrences. I will admit that acceptance of my language is completely dependent on the perceived value of the content.
From a business perspective, I am persuading CLI’s clients to embrace machine learning technologies, and just to be certain that we are paid fairly for the use of the archives, plus an annual fixed royalty payment for the use of the content in potential derivative works. Ultimately, this will be the standard. There is no way publishers will be able to prevent AI from becoming part of all licensing agreements.
I received more than my share of challenges from lawyers who believe that AI use of content must be prevented at any cost. Simply put, while I may agree in part in principle, as stated in my email response above, it is short sighted to think that aggregators and distributors utilizing AI are not going to be the dominant content players in the coming years. The smart publishers will be on board early, with licensing agreements that are consistent with the mission of the publisher, and will be able to dictate what will become the standard terms that all other publishers will have to submit. At CLI, we’re taking the bull by the horns, and are currently working with one of the largest content aggregators in the world to draft contract language for AI utilization of content.
Finally for the many of you that offered congratulations or thanks for the article. You have convinced me to continue the “Confessions” theme in upcoming articles.