Why Eightfold AI requires more careful compliance than a standard ATS
Inferring what candidates didn't tell you
Most ATS tools match explicit skills — a candidate lists Python, the job requires Python, the system matches. Eightfold AI goes further: it infers skills adjacency, career trajectory potential, and likely future performance from patterns in millions of career paths. A candidate who has never listed "project management" may still be ranked highly for a project manager role because Eightfold infers they have the underlying potential.
This is powerful. It is also exactly what the EU AI Act is designed to scrutinise. When a system makes inferences about people that go beyond what they have explicitly shared, the transparency and explainability obligations become harder to meet — because "the AI saw patterns in your career history consistent with future project management success" is not a simple explanation to give a candidate.
The right to explanation
While the EU AI Act does not create an explicit "right to explanation" equivalent to GDPR's Article 22, deployers of high-risk AI must ensure that affected persons can exercise meaningful rights. In practice, if a candidate asks why they were ranked lower than another applicant, you need to be able to provide a substantive answer — not just "the algorithm decided." Work with Eightfold to understand what explanations their system can generate and prepare a process for handling such requests.