thepublishingtimes.comEditorial

The Publishing Times

All the News That’s Fit to Publish

publishing-news · Guides

IFRRO Survey: 92% of Authors and Publishers Say Generative AI Has Hurt Earnings

A study for the international federation of reproduction rights organisations, drawing on a survey of 1,993 respondents, finds near-unanimous doubt that current law protects rightholders from AI training.

The Publishing Times Staff · October 30, 2026 · 4 min read

More than nine in ten authors, publishers and other rightholders who answered a global survey say generative AI has directly hurt their earnings, according to a study released this week by IFRRO, the international federation of reproduction rights organisations. The study argues that large-scale use of protected works to build AI systems without consent, transparency or payment is "neither legally nor economically sustainable."

The survey

The report, The Legal, Policy, and Market Impact of Artificial Intelligence on Text- and Image-Based Works, was written with Professor Eleonora Rosati, a copyright scholar. IFRRO's press release, dated October 28 in Brussels, describes it as combining legal and policy analysis, recent case law, market evidence and the results of IFRRO's own survey.

That survey ran from June 25 to July 24, 2026, and drew 1,993 responses through IFRRO's member organisations. According to the study, 81% of respondents identified as authors, 9% as publishers, 2% as collective management organisations and 8% as other stakeholders. Responses came from every continent, but the United States and the United Kingdom together made up more than 67% of the total, followed by Germany and Canada.

The findings

  • Earnings: A combined 92% strongly agreed or agreed that generative AI had "directly and negatively impacted" rightholders' earnings over the past few years.
  • Displacement: 77% said they were concerned that AI would ultimately displace creative workers.
  • The law: 96% said current law does not adequately protect authors, publishers and other rightholders against unauthorized use of their works by AI developers.
  • Licensing: More than half agreed that AI-specific licences would greatly improve their economic situation.
  • Human work: More than three in four agreed that human-made content will ultimately be preferred to AI-generated content.

In written comments, respondents described unauthorized use of their work for AI training, the flooding of markets with AI-generated content, layoffs and lost revenue. Several also reported a rise in suspicion of AI use that has led to wrongful accusations against human writers. Supporters of licensing stressed that any licences would need careful drafting and enforceable terms if individual authors and smaller rightholders are to benefit.

The respondent pool was drawn from IFRRO's own membership and is weighted toward authors, so the figures reflect the views of rightholders rather than a random sample of the wider book trade.

The argument

Beyond the survey, the study reviews how courts and policymakers have treated "harm" in copyright cases and summarizes market studies that point to revenue displacement across creative sectors. It also notes the growth of a commercial market for licensing content for AI training. The study sets out three conditions it considers fundamental: authors and rightholders must be able to control and consent to the use of their works, established copyright principles must be upheld, and AI systems must be transparent in both training and output.

"This – very timely – IFRRO study ... demonstrates that large-scale unauthorized uses of protected content without consent, remuneration, and transparency raises fundamental legal and economic concerns," Anita Huss-Ekerhult, IFRRO's chief executive and secretary general, said in the release.

IFRRO, based in Brussels, represents more than 160 member organisations in some 90 countries. It said Rosati will present the findings at a public webinar on November 23.

Sources