What does Commission guidelines on prohibited AI practices say about AI in banking?

What are the key points of Commission guidelines on prohibited AI practices?

• Covers each Article 5(1) prohibition with legal analysis and practical examples; explicitly non-binding, with final interpretation reserved for the CJEU.
• Social scoring (Art. 5(1)(c)) is prohibited where AI evaluates people based on social behaviour or personal characteristics and produces detrimental treatment in unrelated contexts or disproportionate to the behaviour; creditworthiness assessment using legitimate financial data is not per se social scoring.
• Exploitation of vulnerabilities (Art. 5(1)(b)) — age, disability, or a specific social or economic situation — is prohibited where it materially distorts behaviour and causes significant harm, a risk area for AI-driven marketing of credit products.
• Manipulative or deceptive techniques (Art. 5(1)(a)) capture subliminal or purposefully deceptive AI design, relevant to customer-facing chatbots and nudging.
• Prohibitions applied from February 2, 2025; violations carry the top penalty tier of up to €35M or 7% of worldwide turnover.
• Emotion recognition in the workplace is banned except for medical or safety reasons — relevant to employee-monitoring tools in banks.

What did Commission guidelines on prohibited AI practices change for banks?

The guidelines are the first Commission statement of how Article 5 will be read, letting banks screen customer-facing and marketing AI against the bans with concrete examples rather than statutory text alone. They confirm that ordinary credit scoring is not social scoring, while flagging vulnerability exploitation and manipulative design as the live risks in retail finance.