The Information Commissioner’s Office has published a list of key considerations for employers who are thinking of procuring an AI tool for recruitment. What are they?
- Risk assessment: Assess the necessity of using AI, complete a data protection impact assessment to identify and mitigate risks, and consult affected groups.
- Explain: Explain AI decisions clearly and honestly to individuals, considering the context and potential impact.
- Data minimising: Collect only the necessary personal data for AI development, ensure accuracy and relevance, and use privacy-preserving techniques.
- Bias and discrimination: Address potential biases early by ensuring data accuracy and representativeness and consulting affected groups.
- Data preparation: Invest time and resources in preparing data, involving multiple human labellers and consulting protected groups for fair outcomes.
- Security: Implement appropriate security measures, conduct regular security risk assessments, and monitor AI systems for anomalies.
- Human review: Ensure meaningful human review of AI decisions, with adequately trained reviewers who can override automated decisions.
- Third-party AI systems: Collaborate with external suppliers to ensure compliance with data protection laws, conduct due diligence, and document roles and responsibilities.
- Legal compliance: AI use must comply with data protection laws, requiring lawful basis for data processing and ensuring transparency and fairness.
- Accountability: Controllers must understand AI systems, explain decisions to individuals, and demonstrate compliance with data protection principles.
