The EU AI Act: What It Means for the People You Lead
Most of the coverage is about fines and risk categories. This guide is about the part that lands on managers: the duty to keep humans in charge, and the duty to make sure your people actually understand the AI they use.
What the Act actually is
The EU AI Act is the first broad law governing how organisations build and use artificial intelligence. It sorts AI systems by the risk they carry, and puts the heaviest duties on systems that affect people directly, including hiring, performance management, promotion, and monitoring at work.
It applies to organisations that use AI affecting people in the EU, not only to those that build it. If you operate in Europe, or plan to, it reaches you. The two duties that fall on everyday leaders, rather than legal teams, are human oversight and AI literacy.
The date that matters
The Act is being phased in over several years. For most employers, the milestone to plan around is 2 December 2027. That is when the rules for high-risk AI systems apply, including many used in hiring, performance and monitoring at work. The EU moved this date from 2 August 2026 in an update that took effect on 27 July 2026. High-risk AI built into regulated products follows on 2 August 2028.
The AI literacy duty has applied since February 2025, and transparency rules since August 2026. Building oversight and literacy into how people work takes months, so the extra time is best used to start now.
The two duties that land on leaders
Human oversight
A person must be able to understand, question, and override the AI when it affects someone. That means a named human is accountable for the decision, not the tool, and they have the authority and the information to step in.
In practice this is a design and management question: who reviews what, when, and how a person can be escalated to. Oversight that exists only on paper will not hold.
AI literacy
The people who use or are affected by AI must understand enough to use it with judgement: what it does well, where it fails, and when to trust it. Employers must take steps to build it. Since the July 2026 update, the Act no longer asks you to guarantee a set level for each person.
Literacy is also what turns adoption around. People use tools they understand and quietly avoid the ones they do not.
Where works councils come in
In Germany and much of Europe, this is not only an AI question. Where a system can monitor staff or shape decisions about them, employee representatives have a legal right to be involved. In Germany, the works council has co-determination rights over technical systems that can monitor performance or behaviour under the Works Constitution Act.
Bringing the works council in early is not a hurdle. It is how you avoid a rollout being halted late, and how you build the trust that adoption depends on. We facilitate these conversations in German and English.
What employers should do now
And in Australia?
Australia has no AI Act. The government has taken a technology-neutral path, with its Guidance for AI Adoption (October 2025) setting out what good practice looks like now. From 10 December 2026, new transparency duties for automated decision-making take effect under changes to the Privacy Act.
The direction of travel is the same in both regions: keep a human accountable, be transparent, and make sure people understand the systems that affect them. Organisations that build this in once tend to meet both sets of expectations.
See where your organisation stands
The AI Change Readiness Diagnostic takes about six minutes and shows you how ready you are across five stages, including oversight and literacy. Or send me an email to talk it through.
This guide is general information, not legal advice. For decisions on your specific obligations, take qualified legal counsel in your jurisdiction.