How can author protect ourselves against increased copyright infringements in this new machine learning world?
Let's begin by first understanding how AI tools are trained. So, to put it simply, machine learning algorithms primarily function by ingesting vast amounts of data and identifying patterns. That means they could have been 'trained' on any publicly available writing of yours. Possibly even copyrighted material (more on that below).
With this in mind, in August (2023), author and publishing industry expert Jane Friedman discovered books falsely attributed to her on Amazon. After reading the first few pages of those books, Friedman immediately suspected they were AI-generated. As Friedman's content is also publicly available online (she's been blogging since 2009), it was also easy enough for someone to ask an AI tool to write something in her specific style, too.
This incident highlighted how AI is making it easier for scammers to use author names and 'voices' to make money. While I'm more AI-curious than AI-negative, the fact is, in today's rapidly advancing digital age, I can't deny we authors have to be more vigilant then ever when it comes to protecting our intellectual property.
I'm not a legal expert. But I wanted to share some of the research I've been doing on this and offer some tips. So buckle up and read on....
First, Be Vigilant
First things first: Make sure you regularly monitor major publishing platforms. If you discover a book or work falsely attributed to your name, act promptly (more on that below). If you're a UK-based author, under the Copyright, Designs and Patents Act 1988 (CDPA), there's a specific provision against "false attribution." So if someone falsely attributes a literary or artistic work to you, both the distributor (like Amazon) and the actual violator (the scammer) can be held liable. While there are services out there that will monitor on your behalf, they're pricey. So seeing as we're all (let's admit it) keen on checking the places where our books live anyway, it shouldn't be too much of a hardship to monitor.
As part of being vigilant, it's also important to constantly educate yourself in this area. The landscape of AI and copyright is always evolving. So staying updated on legal precedents, technological advances, and industry practices via social media (eg. groups like the Savvy Writers' Snug I run on Facebook) and industry press like The Bookseller is a good start.
I would also recommend authors take a look at tools like ChatGPT and get familiar with them. Find out what information they have about you. Experiment with how much they can copy your 'voice'. Even if you hate AI, knowing (what you perceive as) your 'enemy' is smarter, in my opinion, then burying your head in the sand.
Should You Take a Trademark Out?
Some people are now advising taking out trademarks on author names. After all, according to Friedman, the excuse Amazon initially gave for not taking down the books she found was that she hadn't trademarked her name. So there's no doubt trademarking an author name could potentially offer an additional layer of protection against unauthorised use. But trademark registration, especially in the UK, comes with its own set of challenges. It's costly as you need legal assistance, especially global protection which is complex and even more expensive, plus as trademarks are granted for specific classes of goods or services, if you ventured into other areas (like merchandising), other registrations would be necessary. So, while trademarking might offer some protection, it's essential to weigh the benefits against the costs and challenges. Personally? I don't feel it's worth it.
What happens if you discover an infringement?
There's no denying AI tools are making it a lot easier for scammers to fraudulently publish under your name, or produce work under your likeness using a different name... or both. Though Amazon recently took steps to compel disclosure of AI content being uploaded by authors, scammers aren't going to disclose it, are they? And there are currently no ways to easily tell if something has been AI-generated (though those in the know can tell, hence why it's important to familiarise yourself with these tools). So getting these scammy books taken down can, sadly, be a challenge. For example, in Friedman's case, when she went to follow the official steps for reporting copyright infringements on Amazon, she was prompted to link the original book she believed had been scammed. But it wasn’t Friedman's existing work that was being pirated, but rather her name and likeness were being attached to a book without her permission. So this made it even more complicated.
If you discover an obvious infringement on a platform like Amazon (and by obvious, I mean they're clearly using your name and likeness, or have reproduced chunks of your publicly available writing), I recommend starting with official steps.
Here's how I'd approach it:
Gather evidence: Document everything. Screenshots, links, purchase records, and any communication related to the infringement will be vital. It's likely a platform like Amazon will want you to prove you are who you say you are, too. I get it. Imagine someone who has beef with you tries to get your genuine books taken down, claiming to be you? You'd want Amazon to request proof. So get your proof in place as well. Start with screen grabs of your basic, official government-issued identification like a passport or driver's license. In addition, ask your publisher and / or agent to write a headed letter, vouching for your identity and claim to the work. Finally, if you've received royalties for your work, presenting bank statements or financial records can show a monetary trail linked to your authorship can help.
Contact the platform: Notify the platform immediately about the violation. If it's Amazon where you've discovered the issues (where most infringements take place), I'd recommend logging onto Author Central and going via this page: https://author.amazon.com/help/GA9VNYCXXMHGS5VC (keep in mind, links can change all the time. This was correct as of September 2023). There you will find the opportunity to report infringements depending on which country you found them in. If you don't get a response that satisfies you within 48 hours, contact them again. In my experience, they're pretty good at replying within 24 hours. Most platforms have an automated form to fill in now, but if you need a template for contacting a platform, here's a suggestion below:
Subject: Urgent- Unauthorized Sale of Book Using My Name and Likeness OR Copyright Infringement Alert: My Work Being Sold Under Another Name I am writing to bring an urgent matter to your attention. It has come to my attention that a book titled "[Book Title]" [ISBN] is being fraudulently sold on Amazon under my name, [Your Name] and likeness, without my authorisation or consent. OR: My original work titled "[Your Original Work Title]" is being sold under a different name, "[Fraudulent Work Title]". I am a legitimate author. Please find attached proof of my authorship. OR I am the rightful author and copyright holder of [Your Original Work Title]. Enclosed, please find evidence proving my authorship and copyright plus other relevant documentation. I kindly request that you take immediate action to remove this unauthorised listing from your platform to prevent further copyright infringement and damage to my professional reputation. OR I request immediate removal of the infringing material from Amazon's platform to prevent any further violation of my intellectual property rights. I appreciate your immediate attention to this matter and expect a prompt resolution to avoid any legal proceedings. Thank you for your prompt attention to this critical matter.
3. Draw on the community's support: If the platform refuses to take the book/s down, or is irresponsive, use the author and publishing community to put pressure on. Share the issue on social media, contact trade press like The Bookseller, get your agent and publisher to lobby for you (if you have them) and go to an association like Society of Authors or The Alliance of Independent Authors. You may even choose to engage with a solicitor who specialises in copyright law, especially one familiar with the nuances of AI and ML. But this is very costly so do what you can to pile on the pressure before taking such steps.
What About the AI Tools Themselves?
So far, we've mainly covered spammers' use of AI. But what about the AI tools themselves?
Many authors take umbrage to the fact these tools may have been trained on their work. Is it even legal, and even if it is, shouldn't we at least get some credit or compensation (this is what authors like Margaret Atwood and Jodi Picoult are calling for).
First, to answer the legality question. The truth is, it isn't clear. US comedienne Sarah Silverman is suing ChatGPT for copyright infringement. She claims because ChatGPT is able to produce a detailed synopsis of her book, she believes OpenAI, the guys behind ChatGPT, trained its models on pirated versions of her works “without consent, without credit, and without compensation." She isn't alone in this. Other writers are suing too.
These writers are going to face an uphill battle though, particularly given Google's past legal victories over its digital book library. In 2016, the U.S. Supreme Court dismissed claims by authors who argued that Google's scanning and partial displaying of millions of books constituted massive copyright infringement.
It's the same in the UK. There are two clauses in the Copyright, Designs and Patents Act 1988 (CDPA) that can be used by AI tools. You can read the act here but to summarise, it could be argued AI tools like ChatGPT are not infringing copyright because their work comes under 'Making of temporary copies' exception (clause 28A). So, for example, as ChatGPT learns from the web, it supposedly doesn't save exact copies of copyrighted materials. Instead, it learns patterns and information in a generalised manner. This means even if copyrighted content is part of its training data, the model doesn't reproduce or retain specific segments.
However, this become a bit of a grey area when you consider the ability of some ML systems to regurgitate original content. Plus there is also a line in this clause that states temporary copy shouldn't have any money-making value on its own. And yet many AI tools do offer monetised subscription service. But the team behind these tools argue that while they may be part of monetised services, it's the service itself (access to the AI, its computing resources and so on) that's being monetized, not the specific outputs or temporary copies it produces.
Then there's clause (section 29A) of the act. This is the clause that allows schools and universities to reproduce and print off extracts from copyrighted works. It also includes computational analysis, which is key to the learning processes of tools like ChatGPT. The clause states that as long as it's not for commercial research purposes, copyrighted content can be used. The grey area here is ChatGPT does make money from its services with a subscription rate for its newer models. So it's all a bit complicated.
The fact is, this is such a new and fast moving area, the legal world simply hasn't caught up. Many AI-positive people also argue that if we could sue the likes of OpenAI for learning from material, it means author themselves should be sued for reading other novels and 'learning' techniques from our fellow authors (subconsciously or consciously) to improve our craft without providing credit.
This is why thinking about the topic of copyright in AI can give anyone a headache! It's like navigating a maze, and the rules of the game keep changing. But what you can do is be vigilant. Watch what happens with various cases being brought against tool like ChatGPT. Record any concerns. Most of all, understand this new world instead of burying your head.