Anthropic has announced it will watermark all content that is created or edited with Claude.
Claude, similar to ChatGPT, is one of the most widely used AI chatbots in the world.
The move is to comply with a new code in the European Union that asks AI companies to make sure AI-generated content is labelled as such.
It also comes amid increased scrutiny on the impact of AI on creative industries.
A quick editorial note before we continue with this story:
The Daily Aus has a commercial partnership with Anthropic on our newsletters and podcast.
This means they show up in the advertising section of these products.
Partnering with a range of brands keeps our content free for you.
Anthropic has no involvement or influence on our editorial content, including this piece.
Announcement
Now, back to the latest news.
Since 2 August, the European Commission (the executive body of the European Union) has asked AI companies to clearly label AI-generated content to reduce “deception and manipulation”.
The code only applies to Europe.
Your contribution ensures The Daily Aus can continue doing the work you love.
I want to give $10 monthly to help TDA keep doing the news.
Your details:
In response, Anthropic has announced AI-generated content from its systems will be flagged as such with an invisible watermark worldwide.
It said the watermark will carry through even if Claude only proofreads the text.
It also says it will show up even if the content changes after it goes through Claude.
It will also enable Claude users to detect whether an AI-generated watermark is present in the text or files they copy into the chatbot.
Example: A teacher thinks a student used Claude to write an assignment.
They can run it through the chatbot to see if it picks up its watermark.
The changes will immediately come into effect for new models launched in the EU from August, and will eventually be added to previous models.
Bigger picture
There is increasing criticism of the impact of AI on creative industries.
In July, Anthropic paid the largest copyright settlement in history – $US1.5 billion ($AU2.1 billion) – to a group of authors and publishers after it downloaded millions of pirated books to train Claude.
This week, Spotify announced it will label AI-generated artists from mid-September.
It comes after AI songs amassed millions of streams on the platform, with many listeners unaware of the songs’ origins.
