EU AI Act: Common Scenarios
Scenarios
Is my chatbot regulated under the AI Act?
Answer: Limited obligations only
Conditions:
- Must disclose to users they are interacting with AI [Art 50(1)]
- Unless the AI nature is obvious from context
- No other requirements for minimal-risk chatbots
- If chatbot makes high-risk decisions (e.g., credit), higher tier applies
Confidence: High
Citation: Article 50(1)
Can I use AI to screen CVs?
Answer: Yes, but high-risk requirements apply
Conditions:
- Employment AI is high-risk [Annex III(4)]
- Must meet all Art 8-15 requirements
- Must register in EU database
- Must conduct conformity assessment
- Must provide human oversight capability
Confidence: High
Citation: Annex III(4)(a)
Do I need to label AI-generated images?
Answer: Yes
Conditions:
- Must label content as AI-generated [Art 50(4)]
- Label must be machine-readable where technically feasible
- Applies to deepfakes and synthetic media
- Artistic/satirical exceptions exist
Confidence: High
Citation: Article 50(4)
Does this apply to my US company?
Answer: Yes, if serving EU market
Conditions:
- Applies to providers placing AI on EU market regardless of location [Art 2]
- Applies to deployers located in EU
- Applies if AI output is used in EU
- No establishment in EU required to be in scope
Confidence: High
Citation: Article 2
When do different parts come into force?
Answer: Phased implementation
| Date | What applies |
|---|---|
| Feb 2025 | Prohibited practices (Art 5) |
| Aug 2025 | GPAI rules, governance, penalties |
| Aug 2026 | High-risk requirements (Annex III) |
| Aug 2027 | High-risk in regulated products (Annex I) |
Confidence: High
Citation: Article 113