Anthropic Adds Invisible Watermarks to Claude AI Text

An AI company is adding invisible watermarks to its Claude models. This system will track text that is copied and pasted. This move by Anthropic began in early August, following a new ruling under the European Union's AI Act. The EU Commission will now inspect AI models entering the market. The act, effective August 2nd, gives providers until August 2nd, 2027, to comply. The Council of the European Union says this act ensures AI systems are "developed and used responsibly." It addresses risks like bias and discrimination. Anthropic's decision to use watermarks helps them comply with transparency requirements in Article 50 of this new act. These watermarks will apply to users globally, not just in Europe. While not all AI models currently use them, watermarking is expected to become more common as more legislation emerges. This new regulation could set a global standard for AI, similar to GDPR for data privacy.