AI for dynamic freight pricing: what does the AI Act say?

Short answer: AI that calculates freight rates dynamically based on demand, capacity, fuel costs or seasonality is not a standalone category on the high-risk list in Annex III. It is a commercial, business-to-business application. AI literacy obligations apply, and, depending on the deployment, transparency duties. The classification can shift where the system intervenes in decisions about individuals.

Why dynamic freight pricing usually falls outside high-risk

The AI Act (Regulation (EU) 2024/1689) takes a risk-based approach. An AI system is high-risk only if it is on the Annex III list or is a safety component in a regulated product (Annex I). A pricing system that computes freight transport rates between business parties — shippers, forwarders and carriers — does not touch any of the Annex III categories. It makes no decisions about access to essential services for natural persons, consumer creditworthiness, recruitment or employment relationships. For purely commercial B2B pricing the heavy high-risk obligations (risk management system, technical documentation, conformity assessment) therefore do not apply. Note: the AI Act governs product safety and fundamental rights, not competition. Competition rules and the prohibition on price-fixing fall under competition law and continue to apply in full, independently of the AI Act.

When can it shift to high-risk?

The line is drawn by the impact on individuals, not by the name of the application. Annex III lists, among others, AI that assesses the creditworthiness of natural persons or that determines access to essential private services. If a pricing system targets individual consumers and in fact determines their access to, or the terms of, a service, then it must be assessed case by case whether an Annex III category is engaged. Article 5 (prohibited practices) is also relevant: AI may not use manipulative techniques that materially distort people's behaviour. For B2B freight pricing between undertakings these categories typically do not arise, but the actual functioning and the consequences for individuals are decisive, not the product name.

Which obligations apply in any case?

Even if your pricing system is not high-risk, some duties apply. The AI literacy obligation (Article 4) has been in force since 2 February 2025: anyone deploying AI must have a sufficient understanding of how the system works and what its limits are. In addition, the transparency obligation in Article 50 may apply where people interact directly with an AI system or where content is AI-generated; for a price calculation running in the background this is usually not the case, but for a customer-facing chat or quoting bot it is. A conservative approach is to document the actual functions of your system — what data it uses, what decisions it makes and whom those affect — so you can substantiate its classification.

Read more: AI Act: timeline of obligations. Take the scan.