Anthropic says it will watermark text generated by its AI models

Anthropic will watermark textual content generated by its fashions, together with Claude, to adjust to European laws, the corporate now says. The AI mannequin maker confirmed the watermarking in an updated support page.

EU AI Act’s Transparency Code, which took impact on August 2, requires AI corporations to mark AI-generated or edited content material in a method different techniques can establish them.

Anthropic mentioned that each one fashions launched after August 2 will routinely have tech to watermark each computer-generated textual content and information. For information, the corporate is utilizing the C2PA open customary.

The corporate mentioned it’ll prolong help for older fashions as nicely, including that the watermark will journey when customers copy and paste the textual content.

“As a result of the watermark is a part of the textual content, it’ll journey with the textual content when it’s copied and pasted elsewhere, and should persist by way of some enhancing. Watermarking can be utilized on the mannequin stage, which implies it is going to be current regardless of which Claude product or floor the textual content comes from,” the help web page reads.

It’s not clear how a lot enhancing customers must do to take away the watermark. We’ve requested Anthropic to make clear and can replace the story if we hear again.

The corporate famous that watermarking will apply to totally different merchandise like Claude platform API, Claude, Claude Code, Claude Cowork, and Claude Tag.

Platforms at the moment are speeding to watermark AI-generated content material after backlash from customers and to keep away from regulatory scrutiny. Final week, AI music platform Suno said it will mark tracks created on its platform after a spate of authorized challenges. Final month, publication service Substack teamed up with Pangram to flag AI-generated content material. The corporate’s CEO, Chris Greatest, referred to as out Claudefishing, a time period used for folks utilizing AI to generate content material.

Other than Anthropic, different corporations like Black Forest Labs, Google, Meta, Microsoft, OpenAI, and Synthesia have dedicated to adhering to the EU’s code.

Whenever you buy by way of hyperlinks in our articles, we may earn a small commission. This doesn’t have an effect on our editorial independence.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *