Anthropic, one of the leaders in artificial intelligence development, has announced the implementation of special digital watermarks in texts generated by its neural networks, including the popular Claude models. This decision was made as part of complying with European Union requirements to ensure the transparency and safety of AI use. As reported by Techcrunch.com, this step is aimed at full compliance with new rules that came into force on August 2, 2026.
New EU rules: what does the Transparency Code require?
The Transparency Code, which is part of the European Union's Artificial Intelligence Act, obliges AI developers to label content created or edited using computer systems. The goal is to ensure that such materials can be automatically recognized by other systems. This will allow users and regulators to more easily distinguish between content created by humans and that generated by artificial intelligence.
How do Anthropic's digital watermarks work?
Anthropic specialists have developed a technology that embeds watermarks invisible to humans but readable by machines directly into the text. These watermarks are preserved even when the text is copied and transferred to other applications or documents. For now, exact data on how difficult it is to remove a watermark by editing the text remains unknown. However, experts believe that complete removal would require significant changes to the text structure.
Contradictory data
While most sources confirm that Anthropic began implementing watermarks in August 2026, some analysts point to possible delays in the full deployment of the system. According to one source, the process could take several weeks, especially for older versions of Claude models. Other experts believe that watermarks have already been working in test mode since July 2026, but the company officially announced them only after the Transparency Code came into force.
Industry reaction and other companies
The implementation of Anthropic's watermarks is part of a broader trend among tech giants. Last week, the music platform Suno also announced plans to label tracks created with the help of AI. Furthermore, companies such as Google, Meta, Microsoft, OpenAI, Black Forest Labs, and Synthesia have committed to complying with the EU Transparency Code. Experts believe these measures will help reduce the risks of disinformation and increase user trust in digital content.
Prospects and impact on users
The introduction of digital watermarks may change how users interact with AI-generated content. On one hand, this will increase transparency and safety. On the other hand, it may raise questions about privacy and data control. It is too early to judge the long-term consequences, but it is obvious that the market is moving towards greater standardization and regulation of artificial intelligence use.
