Anthropic Will Mark AI-Generated Text with Special Identifiers

Anthropic, one of the leading companies in the artificial intelligence sector, has announced that it will add special digital identifiers to texts generated by its neural networks, including Claude models. According to ixbt.com, the move is intended to ensure full compliance with the European Union’s strict artificial intelligence regulations. Techcrunch.com reports .

The Transparency Code under the European Union’s Artificial Intelligence Act entered into force on August 2 this year. The document requires AI developers to mark computer-generated or edited content using methods that allow other systems to identify it easily. As a result, Anthropic was forced to update its policies.

How does the new technology work?

According to the company’s updated support page, all models released after August 2 will automatically add digital identifiers to texts and files. While the open C2PA standard is used for files, model-level hidden methods are activated for texts.

Experts say these digital identifiers are embedded directly into the text. As a result, even if a user copies the information and moves it elsewhere, the identifier travels with the text and can withstand certain editing processes. It remains unclear how much users would need to alter the text to remove the identifier completely.

Industry-Wide Changes

The digital marking system will be applied to the Claude platform API, the Claude app, as well as various services such as Claude Code, Claude Cowork, and Claude Tag. Anthropic plans to extend this support to older models over time.

Recently, major technology giants and platforms have been rushing to mark AI-generated content to prevent user complaints and potential legal issues. In particular, the Suno music platform announced last week that it would also mark tracks generated in its system.

Google, Meta, Microsoft, OpenAI, Black Forest Labs, and Synthesia have also committed to complying with the European Union’s Transparency Code. Experts believe such initiatives will significantly simplify the process of distinguishing authentic content from AI-generated content in the digital environment in the future.