Summary
Anthropic's new Claude models will implement watermarking for text and AI-generated image tags in response to EU AI regulations, aiming to curb 'AI slop.' However, researchers remain skeptical about the effectiveness of these invisible watermarks.