AI project brings Flaubert to life

A research team led by Alexandre Gefen, a CNRS researcher, has introduced Flaubot, an AI agent designed to let users converse with Gustave Flaubert, a 19th-century French novelist. The project was presented at the Computational Future conference in Lausanne in October 2026 and aims to recreate Flaubert's language using his works, notebooks, and nearly 4,500 letters.

AI responses tied to historical sources

The AI would provide answers linked to archives on Gallica, the digital library of the Bibliothèque nationale de France (BnF), allowing users to check the sources. The project collaborates with the BnF DataLab, which helps researchers use the institution's digital collections.

Challenges in recreating historical style

The project faces challenges, such as accurately reproducing Flaubert's style from a limited corpus and avoiding anachronisms. Gefen stressed that when documents lack answers, Flaubot should clearly state this to maintain user trust.

Funding and future use of the tool

The project needs about 80,000 euros to fund a postdoctoral position, equipment, and training for a Mistral model linked to Gallica. The tool is intended for teachers, researchers, and the public, though the release date is not yet known.

Publishers require AI disclosure

French publisher Gallimard will ask authors to answer "AI or not AI" when submitting manuscripts, starting in January 2027. This follows a controversy involving author Thelyson Orélien and a Canadian publisher. In the US, 85% of book buyers want to know if AI contributed to a book's writing.

Public AI use and concerns

In France, nearly six out of ten people have used AI to write texts, with 83% of those aged 18 to 24 doing so. The use of AI in writing remains controversial, with debates over how much authors should rely on machines.

AI training and copyright issues

In Japan, distributor Nippan admitted to selling books to Anthropic without knowing they were for AI training. This adds context to Anthropic's Panama project. In Europe, 37% of literary translators have considered changing professions due to AI's economic impact.

Transparency calls at Frankfurt Book Fair

At the Frankfurt Book Fair, 120 publishers urged the European Commission and Parliament to ensure copyright protections, calling for transparency about which works trained AI models. In the US, Audible is testing AI interactions with book characters, starting with Dracula.

AI content scrutiny and market concerns

Amazon faces scrutiny over AI-generated content in ebooks, with 20% of fiction titles containing over 25% AI-generated text. The rise of AI content has sparked debates over quality and market saturation, with Amazon claiming to improve its control mechanisms.

AI notices and font tricks

In China, publishers are adding notices to books stating their content cannot be used for AI training, though protections remain hard to enforce. A creative studio and a Danish type foundry developed a font that tricks AI into reading false text, complicating AI training.

Audiobook industry and AI use

Rakuten Kobo expanded its Kobo Plus subscription to include audiobooks in France, while Hachette Livre hired a new executive to boost audiobook production. A study found that 85% of audiobook professionals use AI for tasks like text preparation and voice generation.

Digitization and AI training

The University of Delaware is temporarily removing over 100,000 books for digitization by Google to support AI systems. In the UK, a campaign called We Are Better Than This opposes the unauthorized use of children's authors' works by generative AI.