Indonesian Political, Business & Finance News

OpenAI follows Anthropic in introducing watermarks for content in the European Union

| Source: Antara (Indonesian) Translated from Indonesian | Technology
OpenAI follows Anthropic in introducing watermarks for content in the European Union
Image: Antara (Indonesian)

OpenAI is set to add digital watermarks to both text and code content created within the European Union region. This move is part of OpenAI’s plan to comply with the newly established AI transparency regulations in the EU. According to Engadget, this step follows a similar move previously taken by Anthropic in early 2026.

OpenAI will utilise a proprietary system named textGrain, which adds invisible statistical signals to the model’s word choices. Once applied, detectors can identify these signals to confirm the presence of a watermark.

In simple terms, this watermarking method may not always be perfectly successful in distinguishing between content purely created by AI and that created by humans. OpenAI has acknowledged that shorter texts are more difficult to detect, with a success rate of approximately 80 per cent.

The same applies to content such as mathematics, and even text editing appears to reduce the success rate. OpenAI intends to make this technology available as open source so that others may modify it.

Overall, the watermarking provision will not be activated automatically, except for ChatGPT and Codex text outputs that meet the criteria within the EU. Customers can choose to enable it for specific models, although it has not been specified which models will be available to the general public worldwide.

OpenAI is being somewhat more cautious regarding detection software. Initial access will be limited to “approved researchers and expert organisations”. Furthermore, the detector will not identify users or reveal prompts or conversations.

Article 50 of the EU AI Act requires providers of generative AI systems to ensure that text outputs are identifiable in a machine-readable format. This requirement already applies to new companies entering the EU market, but existing companies have until 2 December to make necessary adjustments and comply.

This means that the adaptation provisions also apply to companies such as OpenAI, Anthropic, Microsoft, Google, and Meta, among others.

More from 6 October 2026 →

Cite this article
Source: jawawa.id (https://jawawa.id/newsitem/openai-follows-anthropic-in-introducing-watermarks-for-content-in-the-european-union-1791270149)
Originally published by Antara (Indonesian) (https://www.antaranews.com/berita/5773077/openai-ikuti-anthropic-berikan-watermark-untuk-konten-di-uni-eropa), translated from Indonesian
Print