World
Anthropic Details Invisible Watermarking in Claude's AI Text
Anthropic has explained the invisible watermarking system for Claude-generated text to meet AI transparency regulations in Europe.
On Friday, Anthropic clarified the functionality of Claude's new watermarking feature, stating that it will utilize a version of SynthID-Text, an open-source technology developed by Google DeepMind. This system is designed to create detectable patterns within the text based on wording probabilities.
The implementation of the watermarking system has faced criticism, highlighting concerns regarding transparency and accountability in AI-generated content. Observers have noted that while the explanation aims to clarify its purpose, it has intensified discussions about the implications of such technology.
The move aligns with European regulations on AI transparency, signaling a continued effort by Anthropic to comply with emerging legal frameworks surrounding artificial intelligence and its applications.