EU AI Act compliance at RenderBob

How the AI Act applies to RenderBob

RenderBob's ordinary Blender rendering service executes user-provided scenes through Blender's Cycles or Eevee engines. Based on its intended purpose and technical operation, we do not consider this ordinary rendering functionality to constitute an AI system under the AI Act. It is also not intended for a high-risk use listed in Article 6 and Annexes I or III. This assessment is separate from the classification of RenderBob's AI-assisted features.

Two RenderBob features do involve AI and are in scope of the AI Act:

• Bob AI, our render-assistant chat and pre-render analysis layer, which uses large language models to interpret scenes, surface issues and answer questions about render configuration.
• BobGen, our generative scene assist feature, which may produce AI-generated images or 3D assets that you can incorporate into your scene.

The Article 50 assessment on this page assumes RenderBob is acting as the provider of Bob AI and BobGen because those systems are offered under RenderBob's name. Provider or deployer status depends on the actual product architecture and contractual allocation of responsibilities. Using an upstream general-purpose AI model does not automatically make RenderBob the provider of that underlying model. RenderBob may nevertheless be the provider of the downstream AI system.

Both Bob AI and BobGen are classified by us as AI systems. The relevant provisions we address below are Article 50(1), 50(2) and 50(5). Article 50(5) governs how information required under paragraphs 1 to 4 must be provided.

Article 50(1), AI system interactions

Obligation

Providers must ensure that AI systems intended to interact directly with natural persons are designed so that the persons concerned are informed they are interacting with an AI system.

How RenderBob meets it

• Bob AI chat is a genuine two-way conversational interface and is likely within Article 50(1). The chat surface is labelled as an AI assistant before any user interaction, and the chat assistant restates its AI nature in its first message of any new conversation.
• One-way warnings, automated estimates, background analysis, and static suggestions — including some BobCheck (pre-render scene analysis) and BobQuote (instant render cost and time estimate) surfaces — may fall outside Article 50(1) under the Commission Guidelines.
• RenderBob voluntarily labels all such surfaces as AI-generated assistance for consistency, using a text tag such as "AI suggestion" rather than colour alone.
• Disclosure for conversational Bob AI surfaces is provided in the user interface at the time of first interaction, in clear, accessible language, taking account of accessibility requirements for vulnerable persons.
• RenderBob's AI disclosures are currently provided in English. Our primary customer base comprises English-speaking professional users in creative and post-production workflows, where English is the standard working language for DCC tools and render pipelines. We will assess demand for translations into other EU official languages as our customer base expands and will publish target dates on this page when we commit to them.

Article 50(2), generative AI output marking

Obligation

Providers of AI systems generating synthetic image, audio, video or text content must mark outputs in a machine-readable format and ensure they are detectable as artificially generated or manipulated.

Article 50(2) expressly concerns synthetic audio, image, video, and text. Standalone 3D files are not expressly listed. RenderBob applies equivalent marking voluntarily where technically appropriate. Whether a final Blender render incorporating an AI-generated asset is itself an output covered by Article 50(2) depends on the system architecture and the nature of the AI contribution.

Article 50(2) does not apply where an AI system performs only an assistive function for standard editing or does not substantially alter the input data. Some Bob AI functions (for example, surfacing render-configuration issues without generating content) fall within this carve-out.

The complete statutory standard requires markings to be:

• machine-readable;
• detectable as artificially generated or manipulated;
• effective;
• interoperable;
• robust;
• reliable, insofar as technically feasible;
• proportionate to content characteristics, implementation costs, and the generally acknowledged state of the art.

How RenderBob meets it

• RenderBob uses C2PA-compatible Content Credentials where supported as part of its approach to machine-readable marking. Depending on the output format and generally acknowledged state of the art, additional or alternative techniques may be required to ensure that markings remain effective, interoperable, robust, reliable, and detectable.
• Where C2PA is not technically supported by an export format, EXIF metadata, PNG text chunks, or sidecar JSON may be used as implementation measures. Metadata can be removed during export, editing, or platform processing. Sidecar JSON alone does not necessarily ensure robust detectability.
• The final EU Code of Practice on Transparency of AI-Generated Content was published on 10 June 2026. The European Commission published its final Guidelines on Article 50 transparency obligations on 20 July 2026. The Code is voluntary, while the obligations in Article 50 are legally binding. RenderBob evaluates its marking and disclosure measures against the final Code and Guidelines.

Article 50(5), manner and timing of disclosures

Article 50(5) applies horizontally to information required under paragraphs 1 to 4. It is not a separate substantive obligation beyond those provisions.

Requirement

Disclosures must be clear and distinguishable, provided no later than the first interaction or exposure, and compliant with applicable accessibility requirements.

How RenderBob applies it

• Information required under Article 50(1) appears in the user interface at the point of first interaction, not buried in our Terms of Service.
• Article 50(2) primarily concerns machine-readable marking and detectability of synthetic outputs, rather than a user-interface disclosure obligation.
• Our public documentation describes each AI feature, the third-party models it uses where relevant, and the type of output it produces.

Application dates

• Article 50 generally applies from 2 August 2026, which is its date of application.
• For systems placed on the market before that date, the Digital Omnibus provides a limited transition until 2 December 2026 for Article 50(2).
• Content created before 2 August 2026 does not generally need to be labelled retroactively, although voluntary labelling is encouraged.
• This transition does not postpone Article 50(1) or the rest of Article 50.

Authoritative sources

• Consolidated Regulation (EU) 2024/1689 (opens in a new tab) — binding legislation
• Official Article 50 text (opens in a new tab)
• Commission Article 50 FAQ (opens in a new tab)
• Final Commission Guidelines (opens in a new tab) — interpretive guidance, not legislation
• Final Code of Practice (opens in a new tab) — voluntary

Beyond Article 50

We have taken additional measures that are not strictly required by Article 50 but support overall AI Act readiness:

• No training on user content. We do not train any AI models on customer scene files, renders or chat content. Your IP stays your IP. This is reflected in our Terms of Service.
• EU data residency. RenderBob compute and storage primarily runs in EU regions. Our Dedicated Fleet offering for B2B customers is single-tenant and EU-resident by default.
• Upstream model documentation. Where Bob AI or BobGen depend on third-party general-purpose AI models, we document the providers used. The current list is available on request from our compliance contact below.
• GDPR alignment. TraynMe OÜ acts as a controller or processor under GDPR depending on the processing activity. Our Data Processing Addendum (DPA) sets out the role applicable to each customer relationship. The DPA is available to customers on request and integrates AI-related processing transparency.
• Incident reporting. We have an internal process for documenting and disclosing serious incidents related to AI features, modelled on Article 73 of the AI Act as a voluntary alignment, even though Article 73 reporting applies formally only to providers of high-risk AI systems.

Ongoing compliance work

We track the following items publicly:

• Publication of an AI Feature Card for each Bob AI and BobGen feature, including upstream model providers, intended use, and known limitations.
• Independent review of our AI Act self-assessment by a qualified EU AI Act practitioner.
• Remaining C2PA Content Credentials coverage across BobGen export paths where the file format supports it.

Compliance contact

For AI Act, GDPR or other compliance questions, contact us at kristjan@renderbob.io or write to:

TraynMe OÜ
Harju maakond, Tallinn, Mustamäe linnaosa, Mäealuse tn 2/4, 12618
Estonia, European Union

For procurement teams: we can complete vendor security and AI Act questionnaires on request. Allow up to one business day for response.

CTA for B2B prospects

For studios and post houses evaluating RenderBob: book a 30-minute compliance call with our team. We will walk through your specific requirements (data residency, model usage, audit trails, DPA terms) and answer questions on the record.

Book a compliance call

Legal notice

This page describes RenderBob's current assessment and implementation approach. It is provided for transparency and does not constitute legal advice or a binding determination by a supervisory authority. The classification of a feature may change as its intended purpose, technical design, or applicable guidance evolves.