AI statement

REVISED 6/12/23 based on feedback

Where We Stand on AI in Publishing

Recent innovations in AI, particularly in art and text generation, narration, and other areas, will have a significant impact on the publishing industry. AI technologies are here to stay and even their developers are stating that there is a need for regulation. If we want to be a part of that conversation, publishing professionals need to be clear about our concerns, expectations, and intentions. We don’t have a formal organization or group that can speak for us, but as individuals we can voice our concerns, attitudes, and hopes. This was created as an open document that others can sign onto or adopt and modify as they need. The point is to speak up.

This version is a near-final draft. Feedback is welcome, but if you are here to discourage or silence, please move along.

We believe that:

Regarding AI & AI-Detection Development

• AI technologies will likely create significant breakthroughs in a wide range of fields, but those gains should be earned through ethical use and acquisition of data.
• the increased speed of progress achieved by acquiring AI training data without consent is not an adequate or legitimate excuse to continue employing those practices.
• the companies and individuals sourcing data for the training of AI technologies should be fully transparent about the copyrighted works used for training, including but not limited to, providing a searchable index of fully-attributed works that includes when, where, and how they were acquired.
• the companies and individuals sourcing data for the training of AI technologies should be required to publicly identify the user-agents of all data scraping tools and bots, provide simple methods to opt-out of collection, and honor robots.txt and other standard bot-blocking technologies.
• copyright holders should have the ability remove their work from training datasets (to prevent further training use) that have been created from digital or analog sources without their explicit consent.
• user input to AI systems/tools should not be stored or used as further training material without opt-in consent.
• AI technologies also have the potential to create significant harm and, to help mitigate some of that damage, the companies producing these tools should be required to provide easily-available, inexpensive (or subsidized), and reliable detection tools.
• detection and detection-avoidance will be locked in a never-ending struggle similar to that seen in computer virus and anti-virus development, but that it is critically important that detection not continue to be downplayed or treated as a lesser priority than the development of new or improved AIs.

 

Regarding the Publishing Industry

• publishers do not have the right to use contracted works in the training of AI or related technologies without contracts that have clauses explicitly granting those rights.
• submitting a work for consideration does not entitle a publisher, agent, or submission software developers to use it in the training of any AI or related technologies.
• publishers or agents that utilize AI in the evaluation of submitted works should indicate that in their submission guidelines.
• publishers or agents should clearly state their position on AI or AI-assisted works in their submission guidelines.
• publishers should make reasonable efforts to prevent third-party use of contracted works as training data for AI technologies.
• publishing contracts should include statements regarding all parties’ use, intended use, or decision not to use AI in the development of the work.
• publishers and booksellers should clearly label works generated by or with the assistance of AI tools as a form of transparency to their customers.
• publishers and agents need to recognize that the present state of detection tools for generated art and text is capable of false positives and false negatives and should not be relied upon as the sole source of determination that a work has been generated.
• publishers should grant authors the right to decline the use of AI-generated or -assisted cover art, translations, or audiobooks/narration on newly, currently, and previously contracted works.

 

Regarding Authors, Artists, Translators, and Other Copyright Holders

• individuals using AI tools should be transparent about its involvement in their process when those works are shared, distributed, or made otherwise available to a third party.
• individuals should be respectful of publishers’, booksellers’, and other professionals’ policies regarding their desire not to work with AI-generated or -assisted works and not misrepresent those works as entirely their own.
• individuals should acknowledge that there are limits to what a publisher can do to prevent the use of their work as training data by a third party that does not respect their right to say “no.”

 

Regarding Governments

• governments should craft meaningful legislation that both protects the rights of individuals, promotes the promise of this technology, and specifies consequences for those who seek to abuse it.
• “fair use” exceptions (and similar exceptions in other countries) with regard to authors’, artists’, translators’, and narrators’ creative output should not apply to the training of AI technologies and that explicit consent to use those works should be required.
• governments should be seeking advice on this legislation from a considerably wider range of people than just those who profit from this technology.
• governments should attempt to reach a mutually-beneficial understanding with other countries on standards regarding AI development and training.
• copyright should not be extended to generated works.

 

 

NOTES:

• AI-generated and -assisted are not specifically defined due to the fact that the technology is evolving and any current definition would likely create future loopholes. Broad terms are meant to encourage transparency and allow the parties involved to determine whether or not such uses cross their individual lines. For example, one’s attitude towards AI-assisted research may be different for non-fiction vs. fiction.
• AI detection is a complex problem on the same level, or perhaps higher, than the development of AI systems themselves. AI developers are in the best position to address these challenges, either within their tools (watermarking, throttling, etc.), via separate applications, or some combination of efforts. They are also the only party capable of collecting sufficient materials generated by their system, should that become a necessary dataset for detection. Industry professionals have been quoted saying things like “get used to it” and have placed insufficient effort into addressing the problems they are creating. Providing a means of detection should be a cost of doing business in AI and not an alternate revenue stream where they can earn from both sides of an AI/AI-detection arms race.
• We are in no way advocating the end of “fair use” or its equivalents in other countries. While members of the tech industry have made it clear that they would like fair use to apply to their AI training data, the courts have not yet agreed with them. We’re saying fair use shouldn’t be permitted to apply in this case. There are already vendors producing “ethically-sourced” datasets and we welcome this development. Preventing the wholesale use of copyrighted works may slow data acquisition (and by extension development of some AI products), but it does not end or undermine it.