In response to the new rules, Anthropic, the company behind Claude AI, is introducing Anthropic AI Watermarks for text created with its AI systems. The invisible watermarks will help the company comply with the EU AI Content Transparency Guidelines, which took effect on August 2, 2026.
Anthropic Adds AI Watermarks to Comply With EU Law
The watermark is not visible to people but can be detected by special systems. According to Anthropic, it stays with the text even after users copy and paste it. The company also says that all new Claude versions launched since August 2 support this type of watermark, highlighting the wider use of technologies such as Tech CEOs AI Clones.
A Watermark Does Not Prove Claude Wrote the Text
Anthropic claims that the Claude watermark has a critical limitation. The presence of a Claude watermark on the document simply proves that Claude edited the text. It does not necessarily imply that Claude generated the original text.
Read more: Alibaba Unveils Its Most Capable AI Model Yet, Close to Moonshot in Scale
For instance, the users may request Claude to perform some functions such as checking spelling, translating text, summarizing text, or converting one file format into another. Under these circumstances, the original concepts or texts can be from another source. Short text or modified text can lack the Claude watermark.
Files Will Get Digital Information Too
Another plan for Anthropics is to include signed metadata in compatible files. For file formats like images, Anthropic is going to apply the C2PA framework. The C2PA framework helps trace the source of digital media and track changes to its metadata.
Anthropic will apply the watermark at the model level, allowing it to remain in Claude outputs across services such as the Claude app, API, Claude Code, and Claude Cowork. The company plans to use the feature worldwide, highlighting growing AI Cybersecurity Legal Responsibility concerns.
Read more: Google Withdraws Google Earth AI Tool Over Misinformation Risks
Anthropic Follows a Wider AI Industry Trend
Anthropic is one of the many tech firms attempting to adhere to the EU’s guidelines on AI transparency. According to the European Commission, the purpose of these regulations is to make it easier for individuals to detect AI-generated or manipulated content.
Google has also created SynthID, an application for labeling AI content. Some other leading companies such as Meta, Microsoft, and OpenAI have backed the EU’s voluntary guidelines regarding AI-generated content. No publicly available watermark detection technology has been provided by Anthropic till now. News source eTimes Pakistan.
