What is the EU AI Act?
On 1 August 2024, the European Union's Artificial Intelligence Act, Regulation (EU) 2024/1689 (“the EU AI Act" or “the act”) entered into force across all 27 EU Member States.
The act puts obligations on companies offering and using AI for certain purposes, including some tasks and use cases in recruitment processes.
Enforcement of the act’s different provisions will happen in phases, as shown here.
Some of the rules under the act are already in force, such as the obligation for companies to train their staff in the use of AI systems. Other rules don’t take effect until months or even years from now. In particular, the rules on high risk AI systems, which we’ll dig deeper into below, don’t take effect until August 2026. And the rules on high risk AI systems actually only apply if, as from that date, those systems are subject to significant changes in their designs.
Now, let’s dig into what the act will bring - in general, and for Teamtailor and our customers in particular!
Whom does the EU AI Act apply to?
Whether a company is subject to the EU AI, and what obligations a company will have, depends on four main factors:
Does the technology used or offered by the company fall under the act’s definition of an ‘AI system’? The act contains a specific definition of an AI System. It needs to be reviewed carefully, so that you adapt to the rules of the act (only) in relation to all systems and services that actually fall under this definition.
What role will the company have when using or offering the technology? The act contains an extensive catalogue of roles, such as provider, deployer, distributor and operator, with varying responsibilities. It will be important to define the role of every company in the “value chain” of creating, further developing and using an AI system.
What level of risk is related to the AI system? The act defines four levels of risks: unacceptable risk, high risk, limited risks and minimal risk.
Does the AI system fall under a specific category of AI systems? For example, the act contains strict obligations on general purpose AI; on general purpose AI systems that can cause a systemic risk; and specific transparency obligations for certain AI Systems.
A large part of the act’s provisions concern high risk AI systems. These provisions are of particular interest when using an AI system in recruitment processes, since the main rule is that any AI system used for recruitment purposes is to be considered a high risk AI system.
But there are exemptions to this rule, for example when the AI system is intended to perform a narrow procedural task, or when it’s intended to improve the result of a previously completed human activity.
What will the EU AI Act mean for recruiting with Teamtailor?
Already today, Teamtailor offers AI-based features in our service. Most of those features are collected in our Co-pilot, which can be used by any customer who chooses to activate them.
The AI technology that is used in our Co-pilot is powered by AI models from Open AI. The GPT AI models are examples of so-called general purpose AI under the EU AI Act, making Open AI a provider of a general purpose AI under the act.
When this technology is used specifically in the context of Co-pilot, it is however Teamtailor who is the provider. This means that Teamtailor needs to assess how the rules of the act will apply to these features.
We have already started our analysis of the AI features that we offer today, based on the four main factors mentioned above.
We believe that some of our AI features will fall under the act’s definition of a high risk AI system, since they can materially influence the outcome of decision making in the recruitment process. When providing these features, Teamtailor will comply with the obligations on providers of high risk AI systems under the act. This includes, for example, keeping detailed documentation of the features; providing instructions of use; establishing risk and quality management systems; performing conformity assessment and post-market surveillance; and registering the AI system in an EU database.
Other features may be subject to the particular transparency requirements under the act, and will be adapted to this, e.g. by ensuring that output is marked as AI-generated where relevant.
However, some of our AI-based features are unlikely to be considered neither high risk nor transparency risk AI. In relation to these features, neither Teamtailor nor the companies using the features will have any particular obligations under the EU AI act.
What’s important to note is that the rules on high risk and transparency risk AI are not yet in force.
And, since we are continuously developing the Teamtailor service, it’s possible that our current features will function and look different than they do today, when the relevant rules of the act become applicable. It’s also likely that other, additional AI-based features are available at that point.
As a Teamtailor customer, you can rest assured that we will be working hard to create any assessments, documentation, information, certifications, registrations or similar that we need to under the act.
And, as part of preparing for the act, we have also summarized the principles that we already now live by in developing and providing AI-based features in these Responsible AI Principles.
What will the EU AI Act mean for Teamtailor’s customers?
Depending on which AI-based features you choose to use in the Teamtailor service, you may have obligations under the EU AI Act, as a so-called deployer.
For example, you may be expected to provide particular information to your candidates about how you use the AI features; ensure there is human oversight of the features, and that the individuals providing this human oversight have the necessary competence, training and authority.
We will be able to provide you with more details on what obligations will likely fall on you, respectively on Teamtailor, by the time that the rules come into force.
If you want to start your preparations for the act already at this point, we recommend that you:
Review the information we provide about our different AI-based features, so that you are well aware of how they can help in your recruitment processes, and how they work.
Consider which of our current AI-based features you already use, or would like to use, and how. For example, will you be relying only on the feature for a particular task, or will you instruct your recruiters to handle certain aspects manually?
In light of this, make your own assessment of how your use of our AI features will influence your candidates’ access to employment. This analysis can be tricky, so you should consult with your legal counsel if you have questions.
This way, you will already have a good picture of your needs and obligations when you receive more detailed analyses and documentation from us, ahead of when the relevant rules come into force.
DISCLAIMER: This text is not legal advice, but created for informative purposes only. We always recommend that you make your own legal assessments and decisions.