Does Article 50 apply to my company? (Probably, yes.)

Here is the one-sentence version: the EU AI Act does not ask where your company is. It asks where your AI system’s output is used.

The test, in plain words

Article 2 of the EU AI Act sets the territorial scope. It covers, among others, providers and deployers of AI systems located in a third country, where the output produced by the AI system is used in the Union. Unpack that for a marketing team:

1. You use an AI system to help produce content — text, images, audio, video. Under the Act, your company is a deployer: you don’t have to build the model, only use one under your authority.
2. That content reaches people in the EU — your blog is readable in Berlin, your newsletter has subscribers in Amsterdam, your LinkedIn posts surface in Paris feeds.
3. Then the output is being “used in the Union,” and the Act’s transparency chapter — Article 50 — is in scope for that content.

No EU office required. No EU revenue threshold. No EU entity. The same extraterritorial design GDPR used — which is exactly why a Californian SaaS company spent 2018 rewriting its cookie banner.

“But we don’t target the EU”

The honest answer: for English-language content published on the open internet, not reaching the EU is the hard thing to argue. Roughly 450 million people, many of whom read English professionally, are inside the scope zone. If your content strategy includes “publish publicly and let search and social distribute it,” EU audiences are in your funnel whether you invited them or not.

There are real edge cases — genuinely internal content, geo-fenced products, non-EU-language local media. If that’s you, this may not be your problem. For everyone else running AI-assisted content in English, the working assumption should be: in scope.

What Article 50 actually asks of you

Article 50 is the transparency chapter — the light-touch regime, not the high-risk one. For AI-assisted content it converges on a few obligations: disclosure where people interact with AI or consume certain AI-generated content, and — the part most coverage skips — an important carve-out. Article 50(4): AI-generated text published to inform the public need not carry a public AI label where it has undergone human review or editorial control and a person holds editorial responsibility for it.

Which means in practice: a real human review step, and a record proving it happened. An unwritten “someone glances at it” policy is not evidence. When a regulator, a client’s procurement team, or your own counsel asks “show me,” you need a documented review point with a name attached — per asset. (Deep dive: our Article 50(4) explainer.)

Why this reaches beyond the EU anyway

Suppose Article 50 somehow didn’t apply to you. The same four pillars — disclosure, human accountability, audit trail, risk-based governance — are independently arriving from other directions:

• US: FTC §5 enforcement orders demand documentation of AI claims and substantiation.
• UK: the CMA’s AI principles emphasise human oversight and accountability.
• Australia: Privacy Act reforms (December 2026) bring automated-decision transparency duties.
• Canada (Quebec): Law 25 requires disclosure and a human review point on automated decisions.
• Singapore: MGAF calls for comprehensive logging of AI-assisted decisions.
• Everywhere: ISO 42001 clause 9.1 and the NIST AI RMF “Measure” function both require evidence of monitoring.

This is why we describe Article 50 as the global compliance ceiling: it is currently the strictest broadly-applicable transparency standard for AI content. Build your workflow to clear it, and the evidence you produce is the same evidence the FTC order, the CMA principle, the Quebec statute, and the ISO auditor each ask for. Build for the ceiling once, answer everyone. (The full argument: the compliance hub.)

The 90-second way to find out where you stand

We built a free Article 50 Readiness Diagnostic — scenario questions mapped to the obligation areas, scored deterministically, with findings written from your own answers. No signup wall. It tells you which obligations plausibly apply to your stack and what evidence you currently could and couldn’t produce.

What CrawlQ does — and honestly, what it doesn’t

CrawlQ Studio is built for Article 50 alignment by design: every AI output can be routed through a named human reviewer before it publishes, and every approval is recorded in a tamper-evident, Merkle-chained audit trail with per-asset inclusion proofs you can export. Live today, opt-in per workspace, EU-hosted.

What we don’t do (yet, or at all): we don’t attach C2PA machine-readable marks (planned); we don’t cover US state-level AI statutes, Canada’s CASL, or the Act’s Annex III high-risk regime; and no tool makes you “compliant” — compliance is a legal state your counsel assesses. What a good tool provides is the working mechanism and the evidence. That’s what we built.

Frequently asked questions

Related reading

• The Article 50 compliance checklist
• Article 50(4): human review instead of an AI label
• The EU AI Act is a 2026 problem, not a 2027 one
• The compliance hub