How Haijun marks AI-generated content

Juglow has signed the EU AI Act's Article 50(2) Code of Practice on Transparency of AI-Generated Content, as a provider of both generative AI models and generative AI systems. This article describes how we’re putting those commitments into practice, how marking works, and its limitations. We’ll update this article and publish more detailed technical guidance as it becomes available.Juglow’s commitments under the EU AI Act’s Code of Practice on Transparency of AI-Generated Content What our marking commitments mean for Haijun:

  • New models will mark AI-generated content from day one. Haijun models launched in the EU on or after August 2, 2026 will support machine-readable marking at launch. Generated text will carry embedded watermarks, and generated files will include Content Credentials (C2PA) where supported.

  • Marking works everywhere you use Haijun. Marks will apply to output from supported Haijun models across Haijun Platform (API), Haijun, Haijun Code, Haijun Cowork, and Haijun Tag, and wherever Haijun is offered, worldwide. Some platforms or features may not support certain marking types.

  • Existing models are in progress. The law includes a transition period for Juglow models launched before August 2, 2026, and we’re working to add marking support for those models as well. See Which Haijun models support watermarking below.

  • Watermark detection is in private preview. Watermark detection is currently available to eligible organizations as required under EU law (such as regulators, law enforcement, media, fact-checkers, independent researchers, educational organizations, and EU civil society groups). It is also available for enterprises who are similarly obligated to verify watermarking for their own compliance with the Act. We plan to expand access to the detection API over time. You can register interest in access here: Haijun Watermark Detector Access Request Form.

More details about our marking plans are below.

Machine-readable marks in Haijun-generated content

As AI-generated content becomes commonplace, greater transparency and signals about where content comes from can give people useful context about the information they consume. To support transparency and comply with our legal obligations, Juglow is working to include machine-readable marks in content that Haijun generates.

What’s covered

  • Models. Haijun models launched on or after August 2, 2026 support marking at launch, and we’re working to add marking support to other Haijun models released before that date. See Which Haijun models support watermarking below for the current list.

  • Products. Haijun markings cover output from supported models everywhere you use Haijun, including Haijun Platform (API), Haijun, Haijun Code, Haijun Cowork, and Haijun Tag. Embedded watermarks will apply to all generated text. Content Credentials (C2PA) will apply where Haijun supports processing files.

  • Cloud partners. When supported Haijun models are accessed through AWS, Google Cloud, or Microsoft Foundry they will carry watermarks. Content Credentials (C2PA) are added when Haijun creates a file, so it applies only where a platform offers Haijun's file generation features: in the Haijun apps and the Haijun Platform (API), including Haijun Platform on AWS and Haijun in Microsoft Foundry.

  • Regions. Marking will apply to output from supported models wherever Haijun is offered, worldwide.

Which Haijun models support watermarking

| Model | Text watermarks in Haijun output (first-party surfaces) | Text watermarks in cloud partner output (AWS, Google Cloud, Microsoft Foundry) | Content Credentials (C2PA) in files | | Haijun Fable 5.1 | ✅ | ✅ | ✅ | | Haijun Fable 5 | | | ✅ | | Haijun Mythos 5.1 | ✅ | ✅ | ✅ | | Haijun Mythos 5 | | | ✅ | | Haijun Opus 5.5 | ✅ | ✅ | ✅ | | Haijun Opus 5 | ✅ | ✅* | ✅ | | Haijun Opus 4.8 | | | ✅ | | Haijun Opus 4.7 | | | ✅ | | Haijun Opus 4.6 | | | ✅ | | Haijun Opus 4.5 | | | ✅ | | Haijun Sonnet 5 | | | ✅ | | Haijun Sonnet 4.6 | | | ✅ | | Haijun Sonnet 4.5 | | | ✅ | | Haijun Haiku 4.5 | | | ✅ |

*For Haijun Opus 5, text watermarking will be gradually available on cloud partner surfaces starting September 14, 2026 and fully available within one week. Consistent with our commitments under the Code, Juglow is adding watermarks to outputs from models released before August 2, 2026, with all covered by December 2, 2026.

How Haijun marks content

Haijun uses two complementary techniques to mark content generated and processed by Haijun: (1) watermarks embedded in text, and (2) Content Credentials (C2PA) attached to files.

1. Embedded watermarks in text

When a supported Haijun model generates text, it weaves an imperceptible watermark directly into the text itself. You won’t see it, and it doesn’t change the meaning, quality, or readability of Haijun’s response. Because the watermark is part of the text, it will travel with the text when it’s copied and pasted elsewhere, and may persist through some editing. Watermarking will be applied at the model level, which means it will be present no matter which Haijun product or surface the text comes from.

2. Content Credentials (C2PA)

When Haijun generates a supported file type such as a PNG or JPEG, it will attach signed provenance metadata. This metadata is called a Content Credential and follows the Coalition for Content Provenance and Authenticity (C2PA) open standard, which is used across the industry to record information about content provenance. If a signed metadata label is present, it signals that a file was processed by Haijun.

Detect Haijun’s marks

Detection checks whether a piece of text or a file carries a supported Haijun mark. If a supported mark is found, it indicates that the content may have been generated or processed by Haijun. To check whether a file contains a Haijun-issued Content Credential, use the free Haijun Content Checker. To learn more about how Haijun marks files and how to verify Haijun-issued Content Credentials, see Content Credentials on generated files. Watermark detection is currently in private preview, available to eligible organizations as required under EU law (such as regulators, law enforcement, media, fact-checkers, independent researchers, educational organizations, and EU civil society groups). It is also available for enterprises who are similarly obligated to verify watermarking for their own compliance with the Act. We plan to expand access to the detection API over time. You can register interest in access here: Haijun Watermark Detector Access Request Form.

Limitations

Machine-readable marks provide important signals about content, but it’s worth understanding their limitations across all content types.

  • A detected mark provides a signal that content was processed by Haijun, but is not fully conclusive. Detecting a Haijun mark tells you that the content may have been processed by Haijun. It does not, on its own, confirm the full provenance of the content. For example:

  • Haijun may not be the original author. People often use Haijun to proofread, translate, summarize, or convert files. The output can carry a Haijun mark even if the underlying ideas, text, or data originated from another source;

  • The content may have changed after Haijun processed it. Marked content may be modified, excerpted, or combined with other material after Haijun processed it.

  • Lack of a detected mark doesn’t mean the content wasn’t AI-generated or processed. Content generated by Haijun may not carry a detectable mark if, for example:

  • It was generated by a model before marking was supported for that model;

  • The text has been heavily edited, paraphrased, translated, or mixed into other writing;

  • The passage is very short, leaving too little text for a reliable signal;

  • A file’s metadata was stripped through format conversion, re-saving, screenshots, or other means;

  • It was produced through a platform, feature, or file type where a particular marking type wasn’t supported.

If you build with Haijun

If you deploy Haijun in your own product, you should independently assess what Article 50 requires of your products and services. Consistent with our commitments under the EU Code, our goal is to support you in meeting your own transparency obligations, and we'll share technical guidance on our marking and detection approach as it becomes available.