dailynewsnblog
Anthropic says it will start digitally watermarking text generated or processed by its artificial intelligence models, including Claude chatbots, and include metadata in other types of files so they can be identified as AI content.
The move is part of Anthropic’s commitments under the European Union’s AI Act, which includes transparency rules that kicked in early this month to ensure users know whether they are interacting with AI or human creations.
“As AI-generated content becomes commonplace, greater transparency and signals about where content comes from can give people useful context about the information they consume,” Anthropic said in an article posted Tuesday on its Claude support website.
“To support transparency and comply with our legal obligations, Anthropic is working to include machine-readable marks in content that Claude generates.”
Anthropic said Claude models launched on or after Aug. 2 — the date the new EU rules took effect — will contain machine-readable marks for text and files generated or processed by Claude.
An “imperceptible” watermark will be woven into the text itself through backend coding, the company explained.
“Because the watermark is part of the text, it will travel with the text when it’s copied and pasted elsewhere, and may persist through some editing,” Anthropic said.
Get daily National news
Get daily Canada news delivered to your inbox so you’ll never miss the day’s top stories.
“Watermarking will be applied at the model level, which means it will be present no matter which Claude product or surface the text comes from.”
AI-generated files such as JPG images will now contain “signed provenance metadata” that shows the file was created and processed by Claude, as well as whether the file has been tampered with.
Anthropic said it is working on detection mechanisms to help users and third parties detect Claude’s watermarks and provenance metadata, and promised to share details in the near future.
The company is working to add the watermarking feature to older Claude models as well.
It warned that the existence of a watermark or metadata “is not fully conclusive” that the content was completely generated by AI — just that it was processed by Claude models.
For example, text originally written by a human but entered into Claude for proofreading, translation or summarizing would carry a watermark after it’s processed. The watermark would also persist after a human modifies the processed text or or combines it with other material.
Anthropic also noted that the “lack of a detected mark doesn’t mean the content wasn’t AI-generated or processed.” Images generated by Claude could be screenshot or otherwise stripped of its metadata through format conversion, for example.
Companies have been increasingly adopting AI transparency rules in recent months while acknowledging users’ concerns.
Anthropic is among over 200 companies — including OpenAI, Meta, Google and Microsoft — that have signed onto the EU’s code of conduct for AI transparency as of this week.
The EU transparency rules include approved labels for AI-generated or -modified content, though companies can also develop their own labels. AI models are required to include watermarking or identifying code under the law or risk hefty fines.
Tuesday also saw Spotify announce that it would clearly label AI-generated artist personas, and will not include those personas in personalized music recommendations.
However, the music streaming platform suggested it’s up to artists themselves to disclose whether components of a song were created with AI by using AI credits, though Spotify does clearly verify artists that are human.
“With AI credits, we have seen that people are eager to take the opportunity to be transparent and share how they used AI as part of their song’s creative process,” Spotify said in a media release. “We’re hoping for a similar level of self disclosure for AI Personas.”
Last month, OpenAI expanded its existing provenance metadata, known as SynthID, in place for images generated by its AI tools like ChatGPT to also cover AI-generated audio.
The company hosts its own verification tool that allows users to check whether content was created by OpenAI tools.
Canada’s federal AI strategy includes transparency as a top priority for building trust in the technology and protecting Canadians.
The federal government has launched public consultations until Sept. 23 on how to develop its own AI transparency rules, with experts warning regulations or changes to legislation are needed fast as the technology quickly evolves.
© 2026 Global News, a division of Corus Entertainment Inc.
