---
title: "AI in Publishing & Marketing – 2026 Survey"
description: "AI adoption is accelerating—across governments, institutions, and the broader economy. But in publishing, the conversation hasn't resolved so much as it's intensified. Concerns about the illegal..."
url: https://www.boxcarmarketing.com/ai-in-publishing-survey/
date: 2026-05-06
modified: 2026-05-11
author: "Monique Sherrett"
type: page
lang: en
---

# AI in Publishing & Marketing – 2026 Survey

AI adoption is accelerating—across governments, institutions, and the broader economy. But in publishing, the conversation hasn’t resolved so much as it’s intensified.

Concerns about the illegal use of copyrighted materials used for AI training purposes, the environmental impact, the quality of content, and the erosion of critical thinking remain very real. In many cases, those concerns are shaping a cautious, even restrictive approach to the use of AI tools. But while the debate continues, workflows are already shifting. Large studies by Section AI ((https://www.sectionai.com/ai/the-ai-proficiency-report)) and Book Industry Study Group ((https://knowledgecenter.bisg.org/25acbpo/)) show that AI use is not hypothetical; it’s happening.

With that in mind, we surveyed Boxcar Marketing’s Underwire subscribers and LinkedIn followers—a group of digitally engaged marketing and sales folks across publishing, education, and the arts—to get a directional snapshot of how this community is actually using AI today. The sample is small, and not statistically significant, but the responses offer something useful nonetheless: a grounded look at how curiosity, caution, and practical adoption are co-existing in real marketing and sales workflows.

- The Boxcar Marketing **survey ran from Apr 2 to May 2, 2026** and was open to Boxcar Marketing subscribers and LinkedIn followers.

- The survey received **24 responses**, which is not a large enough sample to be statistically representative—but **the answers *are* directionally consistent** with larger studies.

The survey responses are below, plus learn more about the (https://www.boxcarmarketing.com/ai-use-in-publishing-marketing/).

!(https://www.boxcarmarketing.com/wp-content/uploads/image-1024x431.png)

> Marketing & Sales: 62.5% of respondents

!(https://www.boxcarmarketing.com/wp-content/uploads/image-1-1024x431.png)

> Non-users (rarely or never): 37.5%
>
>
>
> Regular users (daily or a few times per week): 58.4%
>
>
>
>
> The (https://knowledgecenter.bisg.org/25acbpo/) (run in summer 2025 with 500+ respondents) found that 45.8% of individuals and 48% of organizations use AI. It will be interesting to see the 2026 follow-up results.

!(https://www.boxcarmarketing.com/wp-content/uploads/image-2-1024x487.png)

> Anecdotally, some respondents identified that they work in a Microsoft environment and must stick to Microsoft Copilot. Likewise, others are only allowed the approved Google suite of tools, which includes Gemini.
>
>
>
> Although ChatGPT shows the strongest use (54.2%), Claude’s agentic and productivity tools (Claude Code and Claude Cowork) are quickly gaining attention in publishing circles.

!(https://www.boxcarmarketing.com/wp-content/uploads/image-3-1024x431.png)

> This is worrying. 38.1% report using mostly free versions. Free tiers of ChatGPT and Claude, for example, use inputted data for training models. This is a significant privacy risk as sensitive or proprietary information can be leaked. Free versions also produce lower-quality outputs because they use older models.
>
>
>
> Given the publishing industry’s concerns about inadequate controls on copyrighted materials, the use of free or free + paid versions signals a need for better policies to manage the intellectual property risks associated with AI training data.
>
>
>
> I dug deeper and found that even among daily users:
>
>
>
>
> 42.9% are using a mix of free and paid tools, whereas
>
>
>
> 28.6% are using paid versions or tools are provided by their company.
>
>
>
>
> The BISG report highlighted that only 31% of publishers had a formal AI policy in place at the time of the survey (summer 2025), and only around 30% were using enterprise/closed models—meaning most are using free, unguarded, consumer-facing tools without institutional structure around them (no corporate privacy, security, or retention policies).

!(https://www.boxcarmarketing.com/wp-content/uploads/image-4-1024x431.png)

> 52.4% regularly use AI in their workflows or have gone a step further and created repeatable processes/templates.

!(https://www.boxcarmarketing.com/wp-content/uploads/image-5-1024x487.png)

> Most AI tasks are Marketing & Sales related (writing copy, metadata, keywords, campaign/content ideation, audience or market research).
>
>
>
> A few respondents are using AI for workflow automations and code/script writing.

!(https://www.boxcarmarketing.com/wp-content/uploads/image-6-1024x431.png)

> Primary uses:
>
>
>
>
> Marketing & admin: 30.4%
>
>
>
> Audience or market research: 17.4%
>
>
>
> Campaign or content ideation: 13%
>
>
>
>
> The BISG data confirms that, for publishing organizations, marketing is the top use case, followed by administrative tasks (note-taking, email management), and metadata. And, the workflow-building cohort is small.

!(https://www.boxcarmarketing.com/wp-content/uploads/image-7-1024x431.png)

> Like the larger studies, people are using AI, just not in a way that has a meaningful impact on their work.
>
>
>
>
> Only 21.7% have many valuable use cases
>
>
>
> 47.8% have some
>
>
>
> 30.4% have very few or none

!(https://www.boxcarmarketing.com/wp-content/uploads/image-8-1024x431.png)

> 60% are underway with experimenting, applying, or advanced use.
>
>
>
> 40% are not using AI or are just getting started.
>
>
>
>
> This is supported by the BISG results: only 13.9% of publishing respondents said AI was essential to their work, while the majority are at earlier stages of experimentation. The workflow-building cohort is small.

!(https://www.boxcarmarketing.com/wp-content/uploads/image-9-1024x431.png)

> 33.3% say trust/accuracy concerns are the biggest blocker
>
>
>
> Aside from the 8.3% who cite lack of time, every barrier listed is trust related (accuracy, privacy, internal resistance, lack of clear value). This tells me that despite AI adoption, there is a healthy amount of hesitation, some of which could be handled by demystifying terminology, providing governance and AI policies, and ensuring human agency in the workflows.
>
>
>
> The BISG highlights that over 33% of publisher respondents are ethically opposed to AI. Ethical concerns, trust, and IP are key barriers; and the opposition is actually stronger in the publisher segment than in the full sample.

!(https://www.boxcarmarketing.com/wp-content/uploads/image-10-1024x487.png)

> Upskilling teams on how to effectively (and ethically) use, prompt, and evaluate AI is one of Boxcar Marketing’s goals. And,[ I’m learning too](https://bc-ai.ca/events/2026-10-27-responsible-ai-professional-certification/).

!(https://www.boxcarmarketing.com/wp-content/uploads/image-11-1024x487.png)

> Standardizing documentation seems critical. The challenge is how to best document the various AI models, their different behaviours, their limitations, the use cases across work functions, then to share that information.

Dig deeper into the Boxcar Marketing (https://www.boxcarmarketing.com/ai-use-in-publishing-marketing/) to see how I’m thinking about the right foundations for using AI, and where the most obvious or high-value use cases are.