The EU AI Act and businesses outside Europe

On Sunday 2 August, most of the European Union's AI Act starts to apply. If your business is in Dubai, Riyadh or Bengaluru, it is tempting to treat that as Europe's business. For some companies it is. But the Act reaches beyond the EU's borders, and a few of its rules are simple enough that it makes sense to follow them anyway.

What changed this week

The timetable moved shortly before the deadline. Last November the European Commission proposed delaying parts of the Act, in a package called the Digital Omnibus. The EU institutions agreed it in May, and it entered into force on 27 July.

The main effect is on high-risk AI, the strictest category, covering uses such as recruitment, credit scoring, education and access to essential services. Those obligations, which were due this August, now apply from 2 December 2027. Rules for AI built into regulated products such as medical devices, toys and lifts move to 2 August 2028.

What was not delayed is the part most businesses will actually touch: the transparency rules.

The transparency rules, in plain terms

From 2 August, the Act's transparency obligations apply. The ones that matter for most businesses are these.

• If people interact with an AI system, such as a chatbot, a voice agent or an assistant in an app, they must be told they are dealing with AI, unless it is obvious from the context.
• Companies that provide generative AI systems must mark what those systems produce, such as images, audio, video and text, in a machine-readable way so it can be detected as AI-generated. Systems that were already on the market get until 2 December 2026 to add this.
• If you publish deepfakes, meaning realistic images, audio or video of real people, places or events that were generated or altered by AI, you must say so.
• AI-generated text published to inform the public on matters of public interest must be disclosed, unless a person has reviewed it and someone takes editorial responsibility.

The bans on certain AI uses, such as manipulative techniques and social scoring, have applied since February 2025. So has a duty on organisations to support AI literacy among staff who use AI, in a form the Omnibus has now softened.

Does it apply to you?

The Act covers companies that place AI systems on the EU market, wherever they are based, and providers and deployers outside the EU when the output of their AI system is used in the EU.

In practice, ask yourself three questions:

1. Do people in the EU use our website, app or chatbot, and do we target them by language, currency, delivery or marketing?
2. Do we sell an AI product or feature to customers in the EU?
3. Do we produce AI-generated images, video or audio that are published to audiences in the EU?

If the answers are all no, the Act is mostly something to watch. If any is yes, the transparency rules are worth meeting now. For a Dubai hotel group with European guests, an Indian SaaS company with clients in Germany, or a Gulf retailer shipping to Europe, at least one answer is probably yes.

The short list for most businesses

For a typical business using AI rather than building models, the work is modest.

• Add a clear line at the start of every chatbot and voice agent conversation saying it is an AI assistant. One sentence does it.
• Label AI-generated images and videos used in marketing, especially anything showing realistic people.
• If your AI tools come from a provider, check how they handle machine-readable marking of generated content, since that duty falls mainly on the provider.
• Make a simple list of where you use AI, what for and who supplies it. It takes an afternoon, and you will need it for any future rule in any market.
• Make sure the people on your team who use AI tools have had basic training on what those tools can and cannot be trusted with.

Fines for breaking the transparency rules can reach €15 million or 3% of worldwide annual turnover, whichever is higher.

Why it is worth doing even if you are not covered

Telling people they are talking to an AI is good practice anywhere. Customers anywhere tend to feel misled when they find out late that the helpful agent was a bot. Labelling AI images avoids awkward moments when someone spots one.

Other governments are watching the EU's approach and writing their own rules. A business that already tells people when they are dealing with AI, and knows where it uses AI, will find most new rules easy to meet, wherever they come from.