In a significant move to comply with the European Union’s new AI transparency rules, Anthropic has revealed how it’s watermarking text generated by Claude AI. The company’s text watermarking method will not have easy-to-see visuals, will not be distinguishable to the people who read it, and will not be adding hidden characters to the text. Instead, Anthropic’s method involves leaving a pattern in the text that can only be decoded by someone who has the key for it.

Source: engadget.com

How Claude’s Watermarking Works

Anthropic explained that large language models like Claude pick one word at a time when generating text by choosing from a list of potential appropriate words to use. They pick words randomly, as long as they make sense for the context of what they’re generating. With watermarking on, Claude will use a key to decide on what word to choose instead of using an arbitrary random number generator to pick the next word.

In its example, Anthropic used the digits of pi as a key. The key starts with the digit 2 from the pi sequence 3.1415926535. The next word generated is the sixth in the list of choices, then the fifth, the third, and then the fifth again. This method is an adaptation of Google DeepMind’s SynthID-Text approach to watermarking, which the team described in a paper published in Nature.

Watermarking doesn’t affect the quality of Claude’s output or slow it down, the company said, and it will not require extra tokens or make generations more expensive. The company will release an API that has ‘keys’ to decode Claude’s watermarking and will be able to say whether the AI generated the block of text being checked.

However, Anthropic admits that its text watermarking method does have limitations. It can’t tell whether Claude actually wrote the text or just edited it, which means if you ask the AI to edit something for you, it will be watermarked too. As the company explains, it can only tell that Claude was likely involved with the text at some point. Even translations will be watermarked. If Claude has only proofread and lightly edited the text, or if the text is too short, the watermark may not be enough to be detectable.

Take note that lightly editing Claude-generated text probably won’t remove its watermark. If you want to be sure, you will need to rewrite it completely. The company also mentioned that code has ‘generally less watermarking than some other forms of text’ because it typically requires exact output. If there are no choices to be made in the text generation, then watermarking can’t be applied.

Anthropic will also watermark images by adding a cryptographically signed note in its metadata that says it was generated by Claude. In its announcement, the company said it was applying watermarks to all of Claude’s output at launch because it doesn’t have sure ways to implement the changes by region. The changes will affect output across all Claude products that use models released after August 2. Anthropic will also add watermarking capability to older Claude models over the coming months.

The company’s move to watermark text generated by Claude AI is a significant step towards complying with the EU’s new AI transparency rules. As AI-generated content becomes increasingly prevalent, it’s essential to ensure that users are aware of when they’re interacting with AI-generated text or images.