The Authors Guild is concerned about reports that some publishing professionals are uploading manuscripts and authors’ personal information into consumer-facing AI systems for uses such as generating summaries, assessments, and marketing copy without permission from the authors or adequate guardrails to ensure that the manuscripts are not used by AI companies for training.

Uploading or inputting a copyrighted work or an author’s personal information into AI systems without permission may constitute a violation of the author’s copyright or right of privacy, and it puts the author’s intellectual property and personal information at risk. Editors, agents, and others in the industry who have access to authors’ works should not upload any manuscript to or otherwise prompt consumer-facing chatbots with any author’s works without first getting the author’s written permission. Where consumer-facing chatbots are used in workflows, publishers and other industry professionals should ensure that they opt out of having the work used for training. All of the common chatbots provide this option in their user settings. Publishers should also take care that any internal AI systems are sandboxed models with guardrails to prevent the manuscripts or author information from being used as inputs for training.

Further, to prevent injecting any AI-generated text into an author’s work, publishers should not use AI to edit a manuscript with the exception of basic spelling and grammar- checking applications.

The Authors Guild calls on publishers and others in the industry to enforce strict policies with their staffs to ensure that authors give written permission before their work is uploaded into any AI and to ensure that authors’ works and personal information are not used for training.  

The Authors Guild recommends including the following clause in publishing agreements to prevent objectionable uses.

Publisher shall not upload the Work or any of Author’s personal information to consumer-facing AI systems for purposes such as generating summaries, assessments, or marketing copy without written permission from the author; and when such permission is granted, it shall use guardrails to ensure that the manuscript is not used by AI companies for training.

Publisher agrees that it will not use AI to edit a manuscript, other than for the use of basic spelling and grammar-checking applications. Further, Publisher warrants that any textual or art changes it proposes will not have been created by AI.