OpenAI / Anthropic / Gemini · SaaS

You use OpenAI in your SaaS chatbot. That does not make the AI Act OpenAI’s problem.

Checkbox scanners that map “GPT API → you are only a deployer” skip the product. If you ship the chatbot, you likely still own Article 50 evidence — in force since 2 August 2026.

Reviewed 27 August 2026 · Preliminary technical reference, not legal advice.

Primary sources: EU AI Act (Regulation (EU) 2024/1689) — EUR-Lex · European Commission — AI Act overview · AI Act Service Desk — implementation timeline

Not legal advice. ActBrief prepares a preliminary technical readiness brief for legal review and enterprise diligence — not a conformity assessment, certification, or guarantee of compliance.

The trap

A common SaaS pitch is: pick OpenAI / Anthropic / Google → we tell you that you are a deployer under Art. 3(4) → here is a list. That can be true for some internal uses. It is often wrong for a customer-facing chatbot you place on the market. You determine intended purpose. You design the UX. You may be the provider of that AI system, with the model lab as GPAI provider in the stack.

What a stack-checkbox scan misses

What ActBrief returns instead

  1. Free readiness scan from product URL + purpose — hypotheses with sources and confidence.
  2. Documentation gaps: disclosure, inventory, oversight SOP, lawyer questions.
  3. Optional Article 50 Evidence Pack (€199): touchpoints, copy drafts, screenshot checklist.

Dates that are actually current (August 2026)

If a toolkit still says the Omnibus “has not been adopted” and last-reviewed April 2026, do not use it for deadlines.

Questions this page answers

If my SaaS only calls the OpenAI API, am I just a deployer?
Not automatically. OpenAI may be the GPAI / model provider. If you place your own chatbot or agent on the market and determine its intended purpose, you are often the provider of that AI system — with Article 50 duties that a vendor logo does not discharge. Confirm role with counsel.
Did the Digital Omnibus delay Article 50?
No. The Council approved the Digital Omnibus on AI on 29 June 2026. Article 50 transparency applies from 2 August 2026. Standalone Annex III high-risk duties moved to 2 December 2027. Pages still treating the Omnibus as an unadopted proposal are stale.
Why not a 3-click scanner that only asks which APIs I use?
API logos do not set intended purpose, material influence, or who places the system on the market. ActBrief asks for the product URL and purpose, then returns gaps, missing facts, and lawyer questions — a handoff file, not a vendor checklist.
What is live for OpenAI-powered chatbots right now?
Article 50 transparency where users in the EU interact with AI or see relevant generated content — in force since 2 August 2026, with Commission guidelines on 20 July 2026. Art. 4 literacy has applied since February 2025. High-risk Annex III is a separate 2027 track.
Official sources

Continue in this cluster

Scan the product, not the vendor logo

URL + intended purpose → role hypotheses, Art. 50 gaps, missing facts, lawyer questions. Email optional. Omnibus dates: Council 29 June 2026.

Get a free readiness brief for your product