Swedish readers are giving AI-written books a chilly reception. In Bokbarometern 2026, 65% expressed a negative attitude toward reading or listening to books written wholly or partly with AI. Just over 3% were positive. The enthusiastic camp could hold a comparatively quiet book-club meeting.
The breakdown was 40% very negative, 25% fairly negative, 18% without a position and 14% unsure. These rounded figures describe attitudes, not book purchases or a blind test of literary quality.
A Survey, Not a National Book Ban
The Swedish Publishers Association commissioned Kantar Media to conduct 1,031 online interviews with people aged 18–84 on January 12–16. The methodology summary describes a randomly recruited Sifopanel, with results weighted by gender, age and region.
That makes the findings more useful than a social-media poll answered by whoever happens to be furious that afternoon. It still does not turn a sample into the voice of every Swede, or a question about attitudes into a promise about what each person will buy.
Calling the positive group a “statistical error” is a joke, not an interpretation of the data. A small measured share is still a measured share. We also cannot identify its members as confused respondents, AI entrepreneurs or authors defending their own books. The results do not supply that character list.
The Younger Readers Are Not Simply Shrugging
Among 18–29-year-olds, 72% were negative and just under 5% positive. Younger adults were therefore more represented at both ends, rather than uniformly relaxed about AI authorship.
Using a technology and wanting it involved in a favourite form of art are different preferences. Someone can use an automated tool at work and still care intensely about who wrote a novel. There is no contradiction that requires a generational mystery to solve.
Nor does the result prove that younger people can recognise generated prose on sight. The question supplied the information about AI involvement. Detecting it in an unlabelled book would be a different exercise.
The Human Behind the Sentence
Respondents’ stated objections included valuing a human voice, loyalty to authors and perceived low quality. The full report, by Karl Berglund and Ann Steiner, explores those attitudes in more detail.
That is a useful reminder of what readers may be buying alongside a plot. A book can offer an encounter with another person’s choices: the detail they noticed, the experience they could not leave alone, the sentence they worked on until it sounded right.
This does not establish that every human-written book is good. Anyone who has abandoned a novel halfway through has already completed that peer review. It does suggest that authorship can be part of the attraction, rather than an interchangeable label on the cover.
Our look at the independent-bookstore revival explores another human part of reading culture: discovery and recommendation. That is a related theme, not evidence that the same people answered this survey or that one trend caused the other.
What Writers and Publishers Can Take From It
The broad phrase “wholly or partly” matters. It puts different amounts of AI involvement inside one question. The headline does not tell us whether a respondent would react identically to a generated novel, a generated passage or a much narrower use during revision.
For anyone communicating how a book was made, specificity is more informative than a vague label. What did the tool do? Which creative decisions remained with the author? These are useful questions for trust, without pretending the survey itself creates a legal disclosure requirement.
For writers, the result is encouraging evidence that human authorship matters to many respondents. It is not a guarantee of income, a forecast of future sales or proof that unwanted bulk publishing is disappearing. Those are separate questions with separate evidence.
The most interesting message is also the least futuristic: readers may care about the person on the other side of the page. The cat, which has been demanding individual attention throughout this entire discussion, regards that as an overdue recognition of an obvious principle.
