· Luca Moretti
The AI Act in August 2026: which duties now apply to a small company
Most of the AI Act applies from August 2026. For most small businesses the duty that matters is transparency: say when people talk to an AI.
The EU Artificial Intelligence Act entered into force on 1 August 2024 and phases in: bans on unacceptable uses after six months, rules for general-purpose AI after twelve, most of the rest after 24 months (August 2026), and some high-risk duties after 36.
Four risk levels
- Unacceptable: banned.
- High risk: strict duties (for example CV screening).
- Limited risk: transparency. People must be told they are talking to an AI, and generated or manipulated media must be marked.
- Minimal risk: spam filters, games. Not regulated.
A chatbot on your website or an AI phone assistant sits in “limited risk”: say clearly that it is an AI.
Fines
Up to €35 million or 7% of worldwide turnover for banned uses, €15 million or 3% for other duties, €7.5 million or 1% for misleading information. For SMEs the lower of the two caps applies. A General-Purpose AI Code of Practice was published on 10 July 2025.