Can Anthropic’s invisible watermarks curb ‘AI slop’? Researchers remain sceptical

TL;DR - Anthropic’s new Claude models will watermark generated text and tag generated images in response to EU AI regulations. Researchers remain sceptical that these measures can meaningfully curb low-quality AI-generated content.

• Text outputs will carry invisible watermarks.
• Images will be tagged as AI-generated.
• The changes reflect growing regulatory pressure for identifiable AI content.
• The limited source does not describe the techniques or provide evidence of their effectiveness.