Anthropic's Claud to Add Hidden Watermarks for AI Content Identification

Artificial intelligence (AI) company Anthropic has announced that its AI chatbot Claud will now include hidden watermarks in its generated texts and images that will make it possible to identify AI-generated content.

The company’s move comes in the context of requirements related to the identification and transparency of AI-generated content under the European Union’s AI Act.

According to the company, the technology will be implemented worldwide and will also apply to specific Claud uses (including Claud code).

Hidden watermarks will not normally be visible to humans. However, automated systems will be able to detect that the content was generated by artificial intelligence with the help of specific signs or signals.

The use of watermarking in AI-generated images has already become relatively common. But according to experts, there are some limitations to this technology. For example, resizing the image, changing the format or taking a screenshot of it can affect or remove the watermark in some cases.

Anthropic's decision comes at a time of growing concern over the rapid rise of low-quality content generated by artificial intelligence on the internet. Such mass-produced content is commonly referred to as 'AI slop'.