top of page

Home / Insights / EU AI Act

Paper: Global, if you serve EU customers

 

When the EU AI Act reaches an African company

 

The EU AI Act applies outside Europe. If your AI system is sold into the EU, or its output is used there, the rules can reach you. 2026 amendments moved the main high-risk deadlines back. That buys time to prepare, not a reason to ignore it.

Data Sentinels. Checked against the sources below on 13 November 2026. Not legal advice.

At a glance

 

Law

 

Regulation (EU) 2024/1689, amended by Regulation (EU) 2026/1744

 

Reaches you if

 

You place AI on the EU market, or the output of your AI is used in the EU

 

Already applies

 

Banned practices and AI literacy (2 February 2025); general-purpose AI models (2 August 2025); transparency (2 August 2026)

 

From 2 Dec 2026

 

Two new bans: AI that creates non-consensual intimate images, and AI that creates child sexual abuse material

 

High risk

 

2 December 2027 for uses such as credit scoring, hiring and education; 2 August 2028 for AI built into regulated products

 

Fines

 

Up to EUR 35 million or 7% of worldwide turnover for banned practices; up to EUR 15 million or 3% for most other breaches

How it reaches you

You sell AI into the EU

 

Providers that place AI systems on the EU market are covered wherever they are based (Article 2(1)⁠(a)).

 

Your AI's output is used in the EU

 

Providers and deployers outside the EU are covered when the output of their AI system is used in the EU (Article 2(1)⁠(c)). A score or decision about an EU resident can be enough.

 

Your EU parent or client asks for it

 

Most African firms meet the Act through contracts first: an EU group policy, or a client's supplier questionnaire.

Four levels of risk

 

Some practices are banned. High-risk uses need risk management, data governance, logs, human oversight and registration. Chatbots and generated content need transparency. Everything else is minimal.

 

Chatbots and generated content

 

Transparency duties apply from 2 August 2026. People should know they are talking to AI, and generated content should be marked. Systems already on the market have until 2 December 2026 for marking.

What to do first

 

Three things you can start on Monday.

1. Map which AI touches EU customers or EU markets, including AI inside software you resell.

 

2. Sort each system into a risk level and note the date that applies to it.

 

3. Start the high-risk evidence early: risk management, data governance, logs and human oversight take months to build.

Not sure where you stand?

 

Take the 2-minute self-check. Eight questions, with your grade and red flags on screen.

 

Take the self-check

Want it done for you?

 

Our Data & AI Governance and Risk work maps which of your AI the Act reaches, sorts it by risk and builds the evidence your EU clients will ask for.

 

Use the form below, or book a 30-minute call.

Not ready to talk yet?

 

Get Governance First: one email a month on what changed in the rules we track, and one thing to do about it.

 

Get the newsletter

One call is enough to know if it fits.

 

Send a request and we reply within one working day with times for a call.

 

In the United States? Use our US page.

Prefer email?

 

info@data-sentinels.com

 

Rather pick a time now?

 

Book a 30-minute call. The call covers Data & AI Governance and Risk and the AI Academy.

 

What happens to your details

 

We use them to reply and prepare your call. Calls are recorded for quality and training. Read the privacy notice

What would you like to talk about?
bottom of page