aivalet.nu, a Swedish guide
☰
🇸🇪 SE
🆕 New guide Winningtemp price 2026: how much it costs Read →

Home › Guides › AI Act: checklist

AI Act 2 August 2026: checklist for small businesses

Short answer: On 2 August 2026 the AI Act's transparency requirements begin to apply. For most small businesses this means two things: tell customers when they interact with AI (e.g. a chatbot) and make sure staff understand the tools (AI literacy, a requirement since February 2025). The heavy high-risk requirements are postponed to 2027–2028 and rarely affect small businesses. The checklist below takes 20 minutes, and the template is free, with no registration.
⬇️ Download the AI policy template (Word, free, no email required)

Checklist: 6 points before 2 August

  1. Inventory your AI tools. List everything used at work, including free accounts. (Search the catalogue to see where the data goes.)
  2. Label customer-facing AI. Do you have a chatbot? Add a clear line: "You are chatting with an AI assistant."
  3. Handle AI-generated content. Do you publish AI text or images? Introduce a routine: human review, and labelling where needed.
  4. Talk through AI literacy. One team meeting goes a long way for a small team: what the tools can do, what they guess wrong about, and what must never be fed in. Document that it was done.
  5. Adopt an AI policy. Use the template: fill in the tool list and responsible person, done in 15 minutes.
  6. Check your tools' jurisdiction. Transparency is the legal requirement, but where data ends up is the bigger question. See the CLOUD Act status for your tools.

What probably does NOT apply to you

High-risk requirements (documentation, CE marking, risk management) apply to providers and users of AI in sensitive applications (recruitment, credit decisions, medical devices) and are also postponed to December 2027 (standalone systems) and August 2028 (product-embedded AI). Writing quotes with an AI chat does not subject you to them. Read more in our AI Act guide.

Common questions

What happens on 2 August 2026?

The AI Act transparency requirements (Article 50) start to apply: anyone encountering AI must be informed. Customer-facing chatbots must disclose they are AI, and AI-generated content that is published must be identifiable.

Does the AI Act not apply until 2027?

Both are true. The EU Digital Omnibus (June 2026) postponed HIGH-RISK requirements to December 2027/August 2028, but prohibitions have applied since February 2025 and transparency requirements start on 2 August 2026. The postponement mainly concerns providers of high-risk AI, not ordinary users.

Does a small business need an AI policy?

The law does not require a formal policy, but it requires AI literacy among staff (since February 2025) and transparency (from August 2026). A one-page policy is the simplest way to meet and demonstrate both.

Which tools trigger the transparency requirements?

Mainly AI that interacts directly with humans (customer chatbots) and AI that creates content you publish (text, image, video). Internal aids such as spell-checkers or meeting notes are generally not affected in practice.

What is the risk of ignoring it?

Fines are provided for in the regulation, but for small businesses in 2026 the practical risk is more likely customer trust and contract requirements from larger clients. Supervision is being built up now: PTS coordinates in Sweden, with IMY and Finansinspektionen in their areas.

Sources

General guidance, not legal advice. Updated 14 July 2026. We monitor changes under News.