New standard seeks to give publishers control over AI training data

A coalition of AI developers and technology firms is backing a new standardised approach that would allow publishers and content creators to prevent their work from being used to train artificial intelligence systems. The proposed framework introduces a policy token—a technical mechanism designed to simplify the opt-out process across multiple platforms and services.

Under the current landscape, publishers face a fragmented system where different AI companies employ varying methods for honouring content exclusion requests. The new standard aims to establish a consistent, machine-readable protocol that publishers can implement across their digital properties, signalling their preference not to have their material incorporated into AI training datasets.

Prominent artificial intelligence firms including OpenAI, Anthropic and Google have publicly supported the proposal, indicating industry momentum toward standardisation. The backing of these major players suggests broader industry acceptance may be forthcoming, though the framework has not yet achieved formal standardisation status.

The development reflects growing concerns from media organisations, authors and publishers over the use of their copyrighted content to train increasingly sophisticated AI models. By establishing clearer opt-out mechanisms, the proposed standard could address some of these concerns while potentially reducing legal disputes between content creators and AI developers. However, implementation and universal adoption will determine whether the framework meaningfully protects publisher interests across the sector.