Responsible AI
Part XI governs every use of artificial intelligence within NOVARIC®. Its position is that AI in recruitment touches fundamental rights — access to work — and is therefore regulated activity in substance as well as in law. The framework aligns with the EU AI Act, the OECD AI Principles, ISO/IEC 42001 and the NIST AI Risk Management Framework.
Risk-tier classification
Every AI system is inventoried and classified against EU AI Act risk tiers before deployment. Systems used in recruitment or candidate assessment are treated as high-risk and held to the corresponding control set — irrespective of where a vendor claims the boundary lies.
Mandatory human oversight
No high-risk AI output is final without review by a named, accountable person. No candidate is rejected by an automated system without human review. Oversight is a defined role with documented decisions, not a slogan.
Bias testing
Algorithmic tools undergo bias testing before deployment and on a defined schedule thereafter, across gender, age, origin and other protected characteristics. Systems that fail tolerance thresholds are suspended pending correction.
Data protection and security
AI systems operate within ISO/IEC 27001-aligned information-security management. Candidate data used in any automated process is governed by purpose limitation, minimisation and documented lawful basis.
Transparency
Candidates are informed where AI tools are used in processes affecting them, and may request human review of any significantly automated decision.
Frequently asked questions
How does NOVARIC® comply with the EU AI Act in recruitment?
All AI systems are inventoried and classified against EU AI Act risk tiers. Recruitment-related systems are treated as high-risk, with mandatory human oversight, pre-deployment and scheduled bias testing, and ISO/IEC 42001-aligned management.
Can AI reject a candidate at NOVARIC®?
No. No candidate is rejected by an automated system without review by a named, accountable person. Human oversight of high-risk AI is a non-waivable control under Part XI of the Manual.