EU AI Act Transparency Duties from August 2: Compliance Checks for Cross-border Sellers

Checked on 2026-07-21. Source boundary follows the Chinese release file based on Regulation (EU) 2024/1689 and European Commission press release IP/26/1653 dated July 20, 2026. This article does not promise listing, approval, certification result, traffic, conversion, enforcement outcome or timing.

The EU AI Act transparency obligations are scheduled to apply on August 2, 2026. Cross-border sellers using AI chatbots, AI-generated product images, AI-assisted product descriptions or recommendation systems should identify their AI use cases and prepare disclosure, labeling and record-retention controls before consumer-facing publication.

Why August 2 matters

Regulation (EU) 2024/1689 establishes a risk-based framework for AI systems. The Chinese release summarizes the phased timeline: prohibited AI practices started earlier, transparency obligations are the near-term focus for August 2, 2026, and broader high-risk AI obligations follow later. For sellers, the immediate question is whether AI outputs or AI interactions are visible to EU consumers.

Date	Obligation area	Seller relevance
February 2, 2025	Prohibited AI practices	Baseline risk screen for unacceptable AI uses.
August 2, 2026	Transparency obligations	Disclose AI interactions and label AI-generated content where required.
August 2, 2027	High-risk AI obligations	More detailed controls for systems that fall into high-risk categories.

Three e-commerce AI scenarios to review first

AI customer-service chatbots

If consumers are interacting with an AI system rather than a human representative, the interaction should be disclosed in a clear and timely way.

AI-generated product images and descriptions

Product photos, lifestyle scenes, videos, copy or descriptions generated or substantially modified by AI may need visible labeling. The disclosure should be easy for consumers to notice at the point of viewing, not hidden in long terms.

AI recommendation or profiling tools

If a seller deploys its own recommendation system or consumer-facing AI workflow, it should retain information about system purpose, provider documentation, logs and transparency measures. Platform-native systems may involve different responsibility boundaries, but seller-side records still matter.

Core transparency controls

• List every AI tool used for customer service, product images, product copy, video, review processing or recommendation.
• Classify whether the company is acting as a provider, deployer or both for each workflow.
• Add clear labels for AI-generated or substantially AI-modified content where required.
• Prepare specific disclosure for deepfake-style images, audio or video.
• Keep tool version, creation date, content summary, review record and provider documentation in the product or campaign file.

Seller action checklist

1. Run an AI-use inventory across operations, customer service, design and listing teams.
2. Decide where labels or chatbot disclosures must appear in the customer journey.
3. Build a record-retention folder for AI outputs and the human review decision.
4. Train teams not to publish AI content without the agreed transparency check.

Service boundary

绿色方舟（深圳）认证有限公司 can help sellers map AI use cases, review disclosure points, build record checklists and align marketplace-facing content with the stated evidence boundary. The work is a compliance-path review and document organization service; it is not a legal opinion or regulator-result promise.