DepthThe conclusion and your first steps
Direct answer
How do we set up human oversight of AI?
You describe: You want to know what the AI Act expects of human control over AI decisions: human in, on or over the loop, and who needs to be competent for it. Likely role: deployer (you use the system).
This applies now
- Article 4: AI literacyApplicable
Coming up
- Article 14: human oversightfrom 2 December 2027
- Annex III: high-risk AIfrom 2 December 2027
Human oversight is a core requirement for high-risk AI (Article 14, applicable from 2 December 2027) and already a logical part of the Article 4 measures: oversight only works with people who understand the system, know its limits and dare to intervene. Record per system who oversees, with what mandate and what training.
Your first actions
- Design and assign effective human oversight. Determine oversight measures per system, appoint competent persons and give them the mandate to intervene or stop.
- Justify the Article 6(3) exception against each individual condition. Name which of the four Article 6(3) conditions you invoke, with facts, and separately justify why the system poses no significant risk of harm to health, safety or fundamental rights and does not materially influence the outcome of decision making.
- Take role- and context-specific AI literacy measures. Determine for each role, system and context which combination of instruction, guidance, practice or training is appropriate.
General interpretation, not legal advice. Checked against Regulation (EU) 2024/1689 and the Digital Omnibus (EU) 2026/1744; the official source remains authoritative.
Full map for your situation