AI governance in retail and e-commerce.

Consumer protection law

Recommendations that personalise what products consumers see, and pricing that adjusts based on inferred consumer characteristics, must comply with consumer protection law. Price discrimination based on demographic characteristics inferred from browsing behaviour is unlawful in most jurisdictions.

EU AI Act

AI used in consumer-facing recommendation and personalisation systems generally falls in the limited-risk or minimal-risk category. However, AI customer service chatbots must disclose they are AI, and AI that significantly influences purchasing decisions of vulnerable consumers may attract additional scrutiny.

GDPR and data protection

Personalisation and recommendation AI typically processes significant personal data. Data minimisation, purpose limitation, and the requirement for lawful basis apply. Detailed consumer profiling for targeting purposes requires careful legal basis analysis.

Competition law

Algorithmic pricing coordination — where competing retailers' AI pricing systems converge on similar prices without explicit communication — is an active regulatory concern. Competition authorities in the EU, UK, and US have signalled scrutiny of AI-facilitated pricing coordination.

Dynamic pricing discrimination

Dynamic pricing algorithms that adjust prices based on consumer characteristics — inferred location, device, demographic — have been found by regulators to constitute unlawful price discrimination in several jurisdictions. Intent is not a defence; the discriminatory effect of the algorithm is the focus of enforcement.

Recommendation engine opacity

Consumers who cannot understand why they are shown certain products and prices — and who cannot determine whether they are receiving worse prices or narrower selections than others — face an information asymmetry that regulators are increasingly treating as a consumer protection concern.

Customer service AI making binding commitments

AI chatbots that provide incorrect information about prices, product features, or returns policies can create contractual obligations and consumer law exposure for the retailer, regardless of any disclaimer. Governance must define what AI customer service can and cannot commit to.

AI inventory and demand forecasting failures

Single points of failure in AI demand forecasting — a model trained on pre-pandemic data encountering pandemic-era demand patterns, or a model encountering an unprecedented promotional event — have caused supply chain disruptions that affected customer service and retailer reputation materially.

Has your dynamic pricing AI been assessed for discriminatory pricing patterns — specifically, whether price variations correlate with protected characteristic proxies?

Do your AI customer service systems disclose to users that they are interacting with AI — and what is the escalation pathway to human support?

What limits have you placed on AI customer service authority — and how do you prevent AI from making commitments the business cannot or should not honour?

How do you monitor AI recommendation systems for filter bubble effects or content that could harm vulnerable consumers?

Have you assessed your personalisation practices against GDPR requirements — specifically, the lawful basis for consumer profiling and the right to object?

What monitoring do you have for AI demand forecasting performance — and what is the human escalation process when the model encounters conditions outside its training distribution?