Why Anthropic’s AI watermark is going further than its rivals — for now
Anthropic's AI watermark, which can appear in translations, summaries, and more extensively edited text, has prompted some backlash.
Anthropic CEO Dario Amodei is taking a bold approach to AI transparency by implementing watermarks on AI-generated text that differ from rival companies' methods. These watermarks are designed to remain visible even when text is copied, pasted, translated, or substantially edited. Anthropic's method has sparked controversy, with some Claude users canceling their subscriptions and developers creating tools to remove the watermarks.
The company claims it's introducing the watermark to comply with the EU's AI Act, but some experts argue it goes beyond what is legally required. The EU's Article 50(2) states that AI companies don't need to mark AI outputs when the technology is used for standard editing or when it doesn't substantially change a user's original words or meaning.
However, Anthropic's watermark can still appear in various forms of text manipulation. Critics worry that the watermark could be misconstrued as indicating that Claude wrote the text, even when it only assisted in editing or proofreading. Some view Anthropic's approach as a reflection of its safety-first brand identity and a way to reduce regulatory risks by shifting responsibilities to downstream users and platforms.
As other tech giants like Google, OpenAI, Meta, and xAI also prepare to implement their own AI watermarks, the debate over their implications continues.
Written by urgent.news from Business Insider's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
