Research notes: AI is rewriting the illustrator’s brief
Ivan Massow, 6 November 2026

The report advances the Association of Illustrators’ October survey into the economics and control of publishing commissions. Sources were checked on 6 November. Survey figures are experiences of respondents, not population estimates, percentages of commissions, or event probabilities. The 2025 and 2026 samples are separate. Each percentage uses respondents to the relevant question; practice areas overlap. The AOI is a trade association campaigning for illustrators. Its full report and publisher policy documents are linked in the article.

The anonymous quotation is a permissioned respondent quotation in the original AOI report, not an interview by Tomorrow’s Publisher. Emma Dries’s quotation is from Adam Morgan’s WIRED reporting of 9 October, subsequently updated on 15 October. WIRED’s US staff accounts and company responses are attributed; neither proves practice at a UK operating company.

Bloomsbury’s document is its December 2025 authors-and-illustrators policy. PRH’s August 2026 document is US guidance. Penguin’s current UK/Ireland Cover Design Award rules govern the competition, not every commercial book cover. Policies establish stated boundaries; they do not constitute compliance audits or measurements of savings.

The FT case study was published on 3 October 2026, using May-July 2025 interviews and internal documents. It concerns one organisation. One author, Liz Lohn, worked at the FT when it was conducted and has since left. No causal claim about trust or productivity is drawn from it.

The 2000 AD process account was published on 21 October 2026. It documents Simon Davis’s account of composing and oil-painting a cover; it does not disclose a fee or demonstrate a sales lift.

All fees, hourly values and times in the worked cost example are hypothetical. £20 plus three hours at £60 is £200; two further hours make £320, compared with the assumed £300 commission. No market benchmark is asserted.

Noah, a research and forecasting service, supplied a completed two-component research programme and one bounded post-draft evidence pass. Returned evidence was checked for its actual subject and scope; irrelevant matches and coverage counts were not used as proof of workforce change, public opinion or odds.

Forecast availability: the automatic workflow retained five target definitions and run references but returned no publishable numerical estimates through its final handoff. The returned optional final-calculation skip was used, preserving the existing targets and avoiding additional host estimates or replacement commissions. The terminal response labels each row “Noah final quick forecast unavailable”. No odds are invented or inferred from article counts. The full target register, exact status responses and native run references are retained in the private claim audit. This limitation did not delay publication after the normal factual, image and release checks.

The accompanying cartoon was generated with AI and is labelled accordingly. It is a metaphor for constrained creative direction, not a depiction of a named publisher or evidence of an actual incident.