The short answer
In one paragraph
For a small company that uses AI tools, three things apply today: take measures to support the AI literacy of the staff who use AI (since 2 February 2025), stay away from the banned practices (since the same date), and be transparent when people talk to a chatbot or see deepfake content (since 2 August 2026). The heavier rules for high-risk uses, such as AI that screens job applicants or scores credit, were pushed back by the 2026 Digital Omnibus and now apply from 2 December 2027. The practical first step is an inventory: every AI tool in use, who uses it, on what data, and which risk class it falls in.
By Ian Luca Calcavechia, AI consultant, Kraków · Last updated 2026-11-13
The timeline after the Digital Omnibus
| Date | What applies |
|---|---|
| 1 Aug 2024 | The AI Act (Regulation (EU) 2024/1689) enters into force |
| 2 Feb 2025 | Banned practices; AI literacy measures for providers and deployers (Article 4) |
| 2 Aug 2025 | Obligations for general-purpose AI models (the OpenAI, Anthropic, Google, Mistral level, not their business users) |
| 27 Jul 2026 | The Digital Omnibus on AI, Regulation (EU) 2026/1744, enters into force |
| 2 Aug 2026 | Transparency duties (Article 50): chatbots, deepfakes, some AI-generated text; national authorities start enforcing |
| 2 Dec 2026 | Deadline for machine-readable marking on generative systems already on the market; two new bans (non-consensual intimate imagery and child sexual abuse material) |
| 2 Dec 2027 | High-risk systems listed in Annex III (hiring, education, credit scoring, essential services and others); originally 2 Aug 2026 |
| 2 Aug 2028 | High-risk AI inside regulated products (Annex I: machinery, medical devices, vehicles); originally 2 Aug 2027 |
Sources: AI Act Explorer: Digital Omnibus, ActuIA, October 2026, Gibson Dunn.
What a small company that uses AI must do today
- AI literacy. Article 4 has applied since 2 February 2025. After the Omnibus the wording is softer: providers and deployers “shall take measures to support” AI literacy rather than guarantee a level. In practice: a short training, a written AI use policy, and a record that both exist.
- Transparency. A chatbot on the website or on WhatsApp must make clear it is an AI. Deepfake images, audio or video must be labelled, and AI-generated text published to inform the public on matters of public interest must be disclosed unless a person reviewed it and takes editorial responsibility.
- Banned uses. No manipulation that exploits vulnerabilities, no social scoring, no emotion recognition at work or school (except for medical or safety reasons), no untargeted scraping of faces.
- Know your high-risk uses before December 2027. If AI screens CVs, ranks candidates, monitors workers, scores credit or decides access to essential services, it will be high-risk. Deployers will need human oversight, logs, informed workers and, in some cases, a fundamental rights impact assessment.
Relief for SMEs and small mid-caps
- The Omnibus adds “small mid-cap enterprises” next to SMEs, and both can use a simplified form of technical documentation if they provide AI systems.
- Fines: up to €35 million or 7% of worldwide turnover for banned practices, €15 million or 3% for most other obligations, €7.5 million or 1% for misleading information. For SMEs the lower of the two amounts applies.
- National authorities are still being set up: Germany named the Bundesnetzagentur in July 2026, while France’s law was still pending in October 2026 (ActuIA).
- Companies outside the EU are covered when they place AI on the EU market or when the AI’s output is used in the EU.
A 30-day readiness plan
- Week 1: inventory. List every AI tool and AI feature in use (including the ones inside Office, the CRM and the help desk), who uses it and on what data.
- Week 2: classify. Mark each use as minimal risk, transparency duty, or future high-risk. Flag anything that touches hiring, credit, health or children.
- Week 3: policy and literacy. One page of rules (what may go into which tool, what needs a human check; a free template is here), plus a 60-minute training for the people who use AI. Keep the attendance list.
- Week 4: transparency and vendors. Label the chatbot and AI-generated media, and ask each AI vendor for their AI Act and GDPR documentation.
- Repeat the inventory every 6 months; new AI features arrive inside tools without anyone buying them.
This is practical guidance from an AI consultant, not legal advice. Points that need a legal opinion should go to a lawyer.
Get it done
Start with a fixed package
Questions
Frequently asked
Does the EU AI Act apply if we only use ChatGPT or Copilot?
Usually yes: using an AI system at work makes the company a deployer. For everyday tools that mainly means supporting staff AI literacy, following the transparency rules for anything published, and avoiding the banned practices.
When do the high-risk rules apply after the Digital Omnibus?
High-risk systems listed in Annex III, such as AI used in hiring, education or credit scoring, apply from 2 December 2027. High-risk AI built into regulated products under Annex I applies from 2 August 2028.
Is AI literacy training mandatory?
Article 4 has applied since 2 February 2025 and requires providers and deployers to take measures to support AI literacy. It does not prescribe a course or a certificate; a documented policy and a short training for the people who use AI are the usual evidence.
Do we need an AI officer?
The AI Act does not require a dedicated AI officer. Someone should own the AI inventory and the policy, often the person who already owns GDPR or IT.
We are not based in the EU. Does it apply to us?
It can. The Act covers providers that place AI on the EU market and companies whose AI output is used in the EU, wherever they are based.
More guides
- AI for Medicare brokers: more time with clients, less time on admin
- How much does AI automation cost for a small business in 2026
- n8n vs Make vs Zapier in 2026: which automation tool to choose
- AI for FMOs and IMOs: faster onboarding, cleaner compliance, agents who sell more
- How to hire an AI automation consultant: 10 questions to ask before you pay
- Case study: from a manual, multi-party operation to one CRM with an AI layer
- Case study: from WhatsApp bookings to a self-service platform in about two months
- Case study: 14 back-to-back projects for a shopping-mall group in Saudi Arabia
- AI receptionist for a small business: what it costs in 2026 and what to check first
- Website AI chatbot for a small business: what it costs in 2026 and what to check first
- AI use policy for a small company: a one-page template and how to roll it out
- AI for customer emails in a small business: tools, prices and what to keep with a person
- Financial model and cash flow forecast for a small business: what to include and what it costs
- AI note-takers in meetings: consent, retention and the rules for a small company
- Medicare call recording and AI notes: what the 2027 rules ask of brokers
- TPMO disclaimer and Scope of Appointment: a call checklist for the 2027 rules
- All industries