The EU AI Act now has teeth, what that means if you use AI
Since 2 August 2026 the EU AI Act is enforceable: chatbots and AI assistants must disclose that they are AI, AI-generated content must be marked, national authorities can investigate, and the Commission can fine model providers. Even businesses far from Europe fall in scope when they serve EU customers or build and white-label their own AI systems. It's a transparency checklist, not a panic.
The law is no longer a plan
On 2 August 2026 the EU AI Act moved from text to enforcement. The rules that are now live and enforceable sit mainly in Title IV of the Act, Article 50's transparency duties: a chatbot or AI assistant must disclose it is an AI system, and AI-generated content (deepfakes, synthetic media) must be clearly marked.
That's the operating layer, and the machinery alongside it is now real too. National authorities can investigate deployments, and the European Commission can fine general-purpose AI model providers up to 3% of global turnover or €15M. The long-debated high-risk obligations ride on a later track: under the post-Omnibus timetable they apply from December 2027 and August 2028.
What this actually asks of you
For most businesses the enforceable core is a transparency checklist, not a compliance mountain. It lines up as:
- Disclose AI, if a customer talks to a chatbot, it must be clear they're talking to a system
- Mark AI content, synthetic images, audio and video generated by AI need labelling when they appear in commercial context
- Stay AI-literate, the Act's literacy duty has been watched since February 2025; staff who operate AI need basic competence in its limits
- Know which hat you wear, if you build or white-label an AI system for others to use, you can be treated as a provider with weightier duties
Even if your customers are in Kenya
The Act's scope is about the market in which a system is placed, not where its builder sits. If you serve EU customers, or build and white-label systems that do, the same duties attach to you. African businesses building for European clients are increasingly asked for exactly these basics, disclosure, marking, documentation, in procurements for exactly this reason.
A practical, non-panic checklist
Label the simulator as what it is, mark generated media, and have a one-paragraph policy for how AI is and isn't used in the business. For the 97% of our readers that don't run frontier model infrastructure, the headline is: this Act is now operating, its enforceable heart is transparency, and the professional who cannot honestly document that probably has a problem, the business itself doesn't.
- Where is AI running in your business today? Keep a short inventory
- Is every customer-facing AI clearly disclosed? Clearly mark it
- Who's responsible when an AI answer goes wrong? Name the human
Put this to work in your business
The posts here come from what we see building and running systems every day. If any of it maps to your business, we can help make it operational, AI projects, web platforms, data and security included.
