More than 90% of IFRRO survey respondents say AI has hurt earnings

More than nine in 10 respondents to a global IFRRO survey said generative AI had already harmed their earnings, while 96% judged existing copyright protections inadequate. The findings add a broad rightholder survey to mounting evidence of concern about AI’s economic effects across publishing and other creative industries.

The International Federation of Reproduction Rights Organisations surveyed 1,993 people from June 25 to July 24, including authors, publishers and reproduction rights organisations reached through its membership. The study combines those responses with legal and policy analysis, emerging case law and evidence of market impact.

Respondents identified several ways AI was affecting income: the unauthorised training of large language models on protected work, a flood of AI-generated material in markets, layoffs, lost clients and declining revenues. The study also cited reduced income from zero-click searches and wrongful accusations that creators had used AI.

More than three-quarters of those surveyed worried that AI would ultimately displace creative workers. Nearly all considered the present framework insufficient to protect them, and more than half said AI-specific licences would substantially improve their economic position.

IFRRO said protected works had in numerous instances been used at scale to train AI without rightsholders’ consent or remuneration. Its conclusion was that large-scale unauthorised use without consent, transparency and payment was legally and economically unsustainable. The federation called for rightsholders to have meaningful control over use of their work, stronger copyright enforcement and robust transparency in both model training and output generation.

The study, developed with Professor Dr Eleonora Rosati, also noted that commercial licensing for AI training was already established and growing. IFRRO said research across music, audiovisual and publishing sectors consistently pointed to measurable, sometimes quantified, revenue displacement associated with generative AI.

A separate CEATL survey of 1,785 professional literary translators, launched early this year, reflected similar anxiety within one part of the publishing ecosystem. More than 77% of its respondents had a negative view of generative AI’s development in translation and publishing, while about 30% reported fewer assignments or clients, lost opportunities, or movement into lower-paid work that they attributed to AI. Fifty-five percent opposed use of their work for AI training in any circumstances.

The two surveys measure different groups and should not be treated as a single estimate of creative-sector losses. But both centre consent, remuneration and transparency as the issues facing workers and rightholders as AI use expands.