Tag
The article critiques Anthropic's plan to watermark AI-generated text in Claude models using steganography, which alters the text's meaning and quality, contradicting their claims of imperceptibility.