The debate within the publishing industry regarding artificial intelligence has shifted from peripheral applications like cover design and translations to the core substance of literary works. While contracts previously focused on ensuring human oversight for automated tasks, recent controversies involving prestigious awards have forced a reevaluation of what constitutes original authorship. The turning point arrived with the disqualification of Thélyson Orélien's novel 'C'était ça ou mourir' from the Goncourt Prize list due to AI usage, followed by allegations that David Silva Rojas' winner of the Almadia-Ventosa Arrufat International Narrative Prize was 99% generated by AI. Although the author denies these claims, the incident has triggered an investigation by his Mexican publisher and led the French house Gallimard to announce it will screen all its works using detection software.
In response to this accelerated technological landscape, literary agencies are updating their protocols to ensure transparency. Indent, a New York-based agency representing authors in Spanish, Portuguese, and English, has advised clients to preserve drafts and manually transcribe internet citations. Paula Canal from the agency noted that they are considering contractual clauses requiring authors to disclose any specific use of AI tools, a move being adopted by many US agencies. Similarly, Casanovas & Lynch emphasizes trust but expects authors to report their use of technology. María Lynch highlighted that while some Nobel laureates like Olga Tokarczuk discuss conversing with AI during the writing process, the distinction between using AI as support versus generating entire texts remains legally undefined and unregulated.
Publishers are now integrating explicit clauses into contracts and award guidelines to address these concerns. Pilar Reyes of Penguin Random House stated that authors must explicitly manifest that their work is an original human creation not generated by AI. While detection software like Pangram is being used progressively to screen manuscripts, editors maintain that these tools serve only as support for a broader editorial evaluation, ensuring the human voice remains central. However, challenges remain regarding the cost and effort of screening all submissions, as well as the reliability of detection filters which may become obsolete as technology advances. Ofelia Grande from Siruela suggests the industry may eventually need to determine an acceptable percentage of AI involvement rather than seeking a binary solution.
The issue has also highlighted inconsistencies in literary prize regulations. While many awards lack specific rules on AI, exceptions exist such as the Almadia-Ventosa Arrufat prize and those organized by Grupo Planeta, which explicitly disqualify works partially or totally generated by AI. As the sector grapples with an influx of pseudonymous manuscripts, the conversation is set to continue at the upcoming Formentor Prize meetings in Mallorca, where international editors will gather to discuss these evolving standards collectively.
Sources: El País Cultura.