AI in books and publishing: 12 records so far

Every measured gain and problem on the record for this topic, newest first.Too few to weigh yet. Updated November 10, 2026.

The count12 counted · 7 context

One square per record, darker for stronger evidence:Evidence: 0 high4 moderate8 emerging

12 records so far, too few to weigh yet. Is your study missing? Add it

The records

AI in books and publishing

12 records so far

The records

AI in books and publishing: 12 records so far · Truvace

The count

Gains: 1
Traces: 4
Problems: 7

A Trace is one record with both a gain and a problem, so it counts on both sides: 5 records show a gain and 11 show a problem. 7 context records are in the record list but not counted, because they report no measured gain or problem.

How strong the evidence is

Evidence	Gain side	Problem side
HighStrong study design, backed by at least two independent sources	0	0
ModerateA measured study or an official source	2	4
EmergingEarly, small, unreviewed or reported second-hand	3	7

By type of evidence: 6 news reports; 2 interview studies; 1 benchmark or lab test; 1 regulatory or legal action; 1 investigation by reporters; 1 randomized trial. How evidence is scored

12 records so far, too few to weigh yet. A written summary appears once a topic has at least 20 counted records, with at least 5 on each side. Is your study missing? Add it

Strongest gains

An experiment shows AI ideas made stories more creative; they also made them more alike
In an online randomized experiment, writers given LLM story ideas wrote short stories that evaluators rated more creative, better written and more enjoyable, with the biggest gains for less creative writers.
Randomized controlled trial · Evidence: Moderate · 70/100 · TRV-2026-0386
AI-written stories cluster tightly by model while human writing varies widely, a word-frequency study finds
Word-frequency analysis separated AI-written short stories from human ones overall, though GPT-3.5 and human texts rarely overlapped, so it could help with authorship checks.
Benchmark or lab evaluation · Evidence: Moderate · 52/100 · TRV-2026-0883
AI-written road trip book reads fluently line by line but loses the plot, one study finds
In a qualitative review of selected passages, the 2018 AI-written book 1 the Road produced locally fluent text with occasionally striking word combinations.
Interviews or qualitative study · Evidence: Emerging · 45/100 · TRV-2026-0694

Strongest problems

Book publishers sue Google for copyright infringement over Gemini AI training
Major book publishers allege Google illegally copied millions of copyrighted books supplied for limited services to train its Gemini AI models without permission or payment, harming authors and publishers through competing AI-generated substitutes.
Regulatory or legal action · Evidence: Moderate · 62/100 · TRV-2026-0699
$2m crime novel deal collapses amid questions over AI use
A debut crime novel's reported $2m-plus, 14-way auction collapsed when its agents withdrew it, saying they could no longer authenticate how it was written; AI use was suspected but not proven, and the author denies it.
Investigative reporting · Evidence: Moderate · 55/100 · TRV-2026-0698
India's first AI scam: TVK minister jabs DMK over multi-crore Periyar library fraud
Tamil Nadu's school education minister alleged that thousands of AI-generated books with fabricated authors and factual errors were bought from one publisher for the Periyar library, a claim the DMK rejects.
News report · Evidence: Emerging · 42/100 · TRV-2026-2433