If your organization uses AI to screen CVs, rank applicants or evaluate candidates, the EU AI Act should already be part of your recruitment technology roadmap.
But one important deadline has changed.
The EU’s Digital Omnibus on AI entered into force on 27 July 2026. It moved the application date for the AI Act’s main high-risk requirements for Annex III systems to 2 December 2027.
For employers, that creates more preparation time.
It does not mean AI hiring compliance can wait until 2027.
Why AI Hiring Is Treated Differently
The AI Act identifies certain employment-related AI systems as high-risk because their outputs can materially affect people’s careers and access to employment.
Annex III covers AI intended for recruitment or selection, including systems used to analyse and filter job applications and evaluate candidates. However, Article 6 also provides limited exceptions where an Annex III system does not materially influence decision-making or create significant risks; systems that profile people remain high-risk.
That means an AI hiring platform used to score or evaluate applicants can face very different regulatory requirements from a simple recruitment chatbot.
Organizations need to understand what their system actually does, not simply whether the vendor describes it as “AI.”
What Changed in 2026?
Originally, many high-risk AI requirements were expected to become applicable in August 2026.
Regulation (EU) 2026/1744 changed that timeline.
For high-risk systems covered by Article 6(2) and Annex III — the category that includes many recruitment and candidate-evaluation systems — Sections 1, 2 and 3 of Chapter III now apply from 2 December 2027.
That extra time matters because these requirements cover areas such as risk management, documentation, transparency, human oversight, accuracy and deployer responsibilities.
But Some AI Act Rules Already Apply
December 2027 should not be treated as the date to start thinking about compliance.
AI literacy requirements have applied since February 2025. Following the 2026 amendment, providers and deployers must take measures to support AI literacy among staff and others operating AI systems on their behalf.
Most prohibited AI practices have also applied since February 2025.
There are also transparency obligations relevant to some recruitment tools. Article 50 requires AI systems designed to interact directly with people to inform them that they are interacting with AI unless that is already obvious in context. So an employer using AI interview software should not assume that every obligation has been postponed.
What Will Matter for AI Hiring Platforms?
The current AI Act text gives employers and vendors a clear indication of what they should be preparing for.
Human Oversight
High-risk AI systems must be designed so humans can effectively oversee their operation. The legislation specifically addresses the ability to understand system limitations, interpret outputs, avoid over-reliance on automation and override or disregard an AI result when necessary.
In hiring, that makes the distinction between “AI recommends” and “AI decides” important.
Explainable Candidate Evaluation
Recruiters need enough information to understand how an AI system should be used and interpret its outputs. Article 13 requires high-risk systems to provide information about their intended purpose, performance, limitations and other relevant characteristics.
A score with no supporting evidence will therefore be much harder to govern than a system that lets recruiters inspect what produced that assessment.
Data and Bias Management
For high-risk systems, the Act requires appropriate data-governance practices and examination of potential biases that could negatively affect fundamental rights or lead to prohibited discrimination.
This is particularly relevant when AI is comparing candidates or identifying who proceeds to the next recruitment stage.
Records and Auditability
The Act also includes logging and record-keeping requirements for high-risk systems. Deployers are required, where the relevant provisions apply, to retain automatically generated logs under their control for an appropriate period, normally at least six months unless other law provides otherwise.
Recruitment teams should therefore be asking vendors today: what evidence can we actually retrieve later?
What Employers Can Do Before December 2027
Use the additional time to build the process rather than wait for the deadline:
- map every AI tool currently used in recruitment
- document whether it screens, ranks, recommends or evaluates candidates
- establish clear human review and override procedures
- review what candidate data and evaluation records are retained
- train recruiters to understand the capabilities and limitations of the AI they use
- ask vendors how scoring, logging, transparency and bias monitoring work
These steps also make procurement discussions much more practical. Instead of asking whether a platform is simply “EU AI Act ready,” ask what evidence it can provide for each requirement.
Building Human Review Into AI Hiring
This is also where product design matters.
hSenid Hiring Intelligence allows recruiters to define role-specific interview questions and evaluation keywords, while keeping audio recordings and transcripts available for stakeholder review. hSenid Hiring Intelligence Data…
Its evaluation model also supports CV analysis against job requirements and configurable weighted criteria such as fluency and years of experience. hSenid Hiring Intelligence Data…
Those capabilities can support structured review and auditability, but no feature set by itself proves EU AI Act compliance. Compliance depends on the system’s classification, configuration, intended purpose and how an employer actually deploys it.
Use the Extra Time
December 2027 may feel distant.
For organizations already using AI recruiting, it is not.
The useful question now is not “When does the deadline arrive?”
It is:
“Could we explain today how our AI evaluates a candidate, who reviews its output and what evidence we could produce if that decision were questioned?”
If the answer is unclear, the additional time created by the Digital Omnibus is exactly what organizations should use to fix it.
For organizations looking to standardize screening, automate initial interviews and give recruiters clearer candidate analytics, explore hSenid Hiring Intelligence.
