Anthropic's Claude to Implement Invisible Watermarks for Processed Content

Anthropic has announced that its AI model, Claude, will soon implement machine-readable watermarks for all processed content to comply with the EU's AI Act. This includes both AI-generated and manipulated outputs, with a focus on ensuring compliance globally from the launch of new models. The watermarks will be invisible to users and may apply to content that does not require labeling under the law.