BISG, BookNet Canada, Release Final Survey Report on AI Use in Publishing, Libraries

The U.S.-based Book Industry Study Group (BISG) and BookNet Canada this week released the full report from a summer 2025 survey that explored the use of AI in the English-speaking North American publishing ecosystem.

Overall, the survey, AI Use Across the North American Book Industry: 2025 garnered 559 responses across “a broad range of professional roles,” finding AI use among publishing professionals and librarians in North America to be “uneven and often exploratory,”  marked by a mix of “cautious optimism” and concern about the technology’s development.

“Individuals working at larger organizations report higher levels of AI use and greater interest in training opportunities, while respondents at smaller organizations appear more hesitant,” the report’s executive summary notes. “Across the survey, respondents frequently emphasized concerns about copyright, ethics, job impacts, environmental costs, and the reliability of AI-generated outputs. At the same time, many participants acknowledged that AI tools may offer practical benefits for limited operational tasks when used with human oversight.”

The survey, the preliminary results of which were first discussed in a webinar last fall, highlights a key challenge for the industry, the report states: “balancing experimentation with emerging technologies while developing policies to protect creative labor, intellectual property, and the integrity of published content.”

Efficiency over Creativity

Among the report’s findings, 45.8% of respondents said they use AI as individuals and and 48.0% said their organizations use AI, with usage patterns suggesting that AI is mostly being used to “improve efficiency, streamline workflows, and enhance data-driven decision-making,” rather than to “replace creativity.”

More “sensitive” tasks, such as “content creation and translation, rights-related work, or consumer-facing applications” remain comparatively limited, the survey found.

Among the top areas of AI use in publishing organizations: administrative and operational tasks, and marketing activities top the list, each noted by 29.1% of respondents, followed by data analysis (21.4%); Editorial tasks (19.8%); and metadata and title optimization (16.8%).

Respondents said AI is being used least in rights and licensing management, with just 2.8% of respondents reporting using AI for these tasks, followed by QA testing (4.2%); creating “AI-voiced” audiobooks (5.7%); translation (7.0%); and customer service or “reader engagement” (10.0%).

But while AI adoption is expanding, the report also suggests that “significant concerns” remain for industry professionals, including:

• Inadequate controls around the use of copyrighted material (86.4%)
• Hallucinations (AI providing false or made-up information as fact) (84.3%)
• AI-generated books, including fraudulent or low-quality content, flooding major retail platforms (81.1%)
• Inaccurate, false, or biased training data used by AI systems (79.2%)
• Lack of disclosure to consumers when AI-generated content is used (73.9%)
• Lack of trust in the companies developing and controlling AI technologies (73.7%)
• Legal liability (e.g., copyright infringement, data privacy breaches) (70.3%)
• Reinforcement or amplification of bias, discrimination, or oppression (63.7%)
• Job loss or negative impacts on authors and creators (64.1%)
• Incorrect or misleading content that hinders accessibility (60.7%)
• Job loss or negative impacts on publishing career pathways (57.1%)

Meanwhile, the report also found that “a substantial share” of respondents remain non-users of AI, “underscoring that adoption is far from universal.”

Responses from the Canadian and U.S markets were “broadly similar,” the report notes, suggesting that the two markets “are moving at a similar pace.”

However, some differences did emerge from the survey data. For example, among the differences, a higher proportion of Canadian respondents (45.7%) indicated that more ethical guidance around AI adoption would be useful compared to U.S. respondents (34.6%); and Canadian respondents (43.5%) were more likely than their American counterparts (31.9%) to identify sustainability “as a pain point related to AI use.”

Libraries vs. Publishers

Also notable, AI use was relatively similar between publishers and librarians, with 43.7% of individuals working at publishers reporting the use of AI tools in their work compared to 42.5% of library workers.

But there were also some notable differences between the two sectors, particularly in terms of how AI is being used, with the data suggesting publishers are primarily leveraging AI “to support outward-facing functions like marketing,” while libraries are looking to AI more for help with “internal workflows and information-management” processes.

“There are some fascinating implications that emerge when looking across findings,” the report states. “As seen in the data, publishers seem to be using AI as a megaphone to produce more marketing and metadata at scale, while libraries are becoming quality gatekeepers increasingly concerned about AI-generated content flooding the ecosystem,” the report concludes. “This dynamic creates a growing tension between efficiency and administrative burden: as AI accelerates content creation in one part of the ecosystem, it shifts the responsibility for verification and quality control onto others.”

In addition to the survey questions, respondents also shared 170 additional comments from respondents—and the sentiments were predominantly critical: about 72% were negative, 20% neutral, and only 8% positive.

“AI has no place in the book industry, or in any industry for that matter, but especially not in one that relies on human creativity,” one commenter noted. “AI is as likely to destroy as it is to disrupt the publishing industry,” noted another.

Another commenter remarked on the different attitudes toward the technology among departments: “Our editorial team members are highly suspicious and anxious about the dehumanizing impacts. Our marketing team members are highly excited about AI tools and their applications for simplifying their work. We have yet to resolve these sorts of tensions.”

Others questioned the value of the technology when weighed against the potential negative impacts.

“Tasks that might have taken days can take moments. However, I do fear that the costs of AI, as an industry and a society, vastly outweigh these marginal shortcuts,” the comment reads. “We need way more protections for jobs and authors/creators,” another said.

The report concludes by urging more research to help the industry better understand AI’s evolving role.

“Nearly half the industry is using AI, yet qualitative sentiment is overwhelmingly negative. The industry is adopting AI faster than it is becoming comfortable with it,” the report concludes. “By approaching AI with intention, the industry has the opportunity to make informed decisions about its adoption or lack thereof and, if sensible, harness its potential benefits while protecting the creative, professional, and cultural values that underpin the publishing ecosystem.”

The preliminary results of this survey were discussed a webinar back in September 2025, and BISG/Booknet Canada will host another free webinar titled: AI in Publishing: Editorial, Metadata, and Reader Trust on Thursday, May 7th at 1 p.m. ET, which will address the survey results.

You can download a PDF of the full report here.

A version of this article also appears in Publishing Perspectives.