Which AI use cases in retail and ecommerce are high risk under the EU AI Act?

Scoring individual agents' interactions to monitor and evaluate their performance and behavior falls under Annex III point 4(b), employment and worker management. The Article 6(3) exception does not apply where the system profiles natural persons. Inferring agents' emotions is prohibited under Article 5(1)(f), except for medical or safety reasons. Inferring customers' emotions from their voice is emotion recognition on biometric data: high risk under Annex III point 1(c), and Article 50(3) requires informing the people exposed to it. Analytics that only aggregate interaction themes without evaluating individuals can fall outside the high risk category.

Writing product content and extracting catalog attributes is not an Annex III use and makes no decisions about people. When a retailer uses a third party generator, the use is minimal risk for the retailer: the Article 50(2) duty to mark generated text in a machine readable way falls on the provider of that system. When a retailer builds and operates its own generating system and puts it into service under its own name, it is the provider and must mark the output, unless the exception for systems that only assist standard editing or do not substantially alter the input applies. Article 50(4) covers text published to inform the public on matters of public interest, not product listings. Consumer protection law applies to what the listing says in every case.

Feeds and endpoints alone are minimal risk. A merchant that provides its own conversational agent (its own LLM app or assistant, inside an AI platform or on its channels) takes the Article 50(1) transparency duties for that system: people must be told they are interacting with an AI system, unless that is obvious from the circumstances. Where the shopper talks to a third party assistant such as ChatGPT, the assistant is run by its provider, who carries its own obligations.

A system that compares the items a shopper handles with the items scanned does not appear in Annex III and carries minimal risk, though the retailer owes AI literacy under Article 4. The tier changes with the design: for a retailer, facial recognition or other biometric identification of shoppers, or inferring their emotions, falls under Annex III point 1 and is high risk, with Article 50(3) transparency duties for emotion recognition. Inferring emotions of store staff would fall under the Article 5(1)(f) prohibition. A design that identifies or profiles people needs a separate legal assessment.

A size advisor is not listed in Annex III, and a size widget on a product page is minimal risk. The design decides the rest. Article 50 transparency applies only if the size advice is delivered through a conversational assistant that interacts directly with shoppers, and then they must be told it is AI unless that is obvious from the context. Body measurements derived from photos can be biometric data (Article 3(34)). Photo based measurement is likely not a biometric categorization system under Article 3(40) while it serves only as an ancillary feature of the sale, for example previewing a product on the shopper, as recital 16 says of body filters on online marketplaces. It would leave that exception if the system inferred sensitive characteristics from the images or reused them for another purpose.

Screening for a weapon shape, or searching footage by a described attribute such as clothing, is not itself the remote biometric identification or categorization of a natural person, as long as it flags an object or an attribute rather than matching or categorizing an individual. It becomes high risk under Annex III point 1(a) if the system performs remote biometric identification, whether real time or after the fact and in any setting, for example matching faces against a watchlist, and under Annex III point 1(b) if it performs biometric categorization. It is high risk under Annex III point 6 if used by or on behalf of law enforcement for profiling a natural person. Real time remote biometric identification by or on behalf of law enforcement in a publicly accessible space is prohibited outright under Article 5(1)(h), subject only to its narrow, listed exceptions.

An assistant that only answers product, process and inventory questions for employees is not listed in Annex III and is not a high risk system, with AI literacy duties for the deployer under Article 4. Article 50(1) transparency duties also reach systems that interact directly with people, employees included, unless it is obvious from the circumstances that they are dealing with AI. Annex III point 4(b) covers systems intended to allocate tasks based on individual behavior or personal traits or to monitor and evaluate worker performance and behavior. An associate app that also assigns tasks to individual associates or scores them falls to be assessed under that point, so the design decides the tier. Walmart's separate AI task tool, which recommends and prioritizes tasks for overnight stocking, shows how close the two categories sit, and the source does not say whether it allocates by individual behavior.

Translating and adapting commercial content is not an Annex III use and makes no decisions about people. When an organization uses a third party translation or generation tool, the use is minimal risk for the organization: the Article 50(2) duty to mark generated text in a machine readable way falls on the provider of that system, and beyond AI literacy no specific deployer obligations apply. When an organization builds and operates its own generating system and puts it into service under its own name, it is the provider and must mark the output, unless the exception for systems that only assist standard editing or do not substantially alter the input or its semantics applies. A faithful translation may fall within that exception; transcreation that rewrites the message for a market alters the semantics and is less likely to. Article 50(4) covers deepfakes and text published to inform the public on matters of public interest, not marketing translations. Consumer protection and advertising rules apply to the translated text as to the original.

Deciding a loyalty offer or reward for a customer is not listed in Annex III and does not itself evaluate creditworthiness, employment or access to an essential service, so a back office personalization engine is minimal risk. If a conversational channel is added to talk to members directly, Article 50 would apply and members must be told they are dealing with AI. Personalization that deploys manipulative or deceptive techniques, or exploits a member's vulnerabilities due to age, disability or a specific social or economic situation, in a way that causes or is reasonably likely to cause significant harm, is prohibited under Article 5(1)(a) and (b).

Most personalization and content generation is minimal risk. Providers of systems that generate synthetic audio, image, video or text content must mark the output as artificially generated, and deployers must disclose deep fakes (Article 50(2) and 50(4)). Personalization that deploys manipulative or deceptive techniques, or exploits vulnerabilities due to age, disability or a specific social or economic situation, in a way that causes or is reasonably likely to cause significant harm, is prohibited under Article 5(1)(a) and (b). Using AI to assess creditworthiness or to price life and health insurance is high risk under Annex III point 5(b) and 5(c) and belongs on its own page. Outside the AI Act, the FCA Consumer Duty applies only to FCA regulated firms (the financial services slice of this use case), and the Telephone Consumer Protection Act applies only to campaigns delivered by call or text message in the US.

Annex III point 4 of the EU AI Act covers employment and worker management. Point 4(b) lists AI systems intended to be used to make decisions affecting terms of work related relationships, the promotion or termination of work related contractual relationships, to allocate tasks based on individual behavior or personal traits or characteristics, or to monitor and evaluate a worker's performance and behavior, as high risk. A scheduling system that allocates shifts based on an employee's own sales or productivity data falls within that category and needs the risk management, data governance, logging, human oversight and other obligations the Act sets for high risk systems. A system that schedules only from demand forecasts and each employee's stated availability is less likely to fall under Annex III, but a system whose schedules decide hours or other terms of work can still be caught by point 4(b), so the design, and any Article 6(3) exemption, needs assessing case by case.

Classifying and summarizing text feedback is minimal risk. The tier changes if the system infers emotions from customers' voices or faces in calls or video: emotion recognition based on biometric data is listed as high risk in Annex III point 1(c) and triggers the Article 50(3) duty to inform the people exposed. Analyzing feedback from employees to evaluate individual workers moves it towards Annex III point 4(b), and emotion recognition in the workplace is prohibited by Article 5(1)(f), except for medical or safety reasons.

Planning routes for a fleet is not an Annex III use. It becomes high risk under Annex III point 4(b) when the system allocates tasks to drivers on the basis of individual behavior or personal traits, or monitors and evaluates their performance and behavior, for example by assigning harder routes or scoring drivers from telematics and using that for pay or discipline. A system that only optimizes stops, vehicles and loads, with no per driver profiling, is usually minimal risk.

A back office agent that turns business customers' order messages into draft sales orders does not decide on access to services, credit or employment, which is minimal risk. Article 50(1) applies to providers of AI systems intended to interact directly with natural persons. Reading inbound orders and sending a one way acknowledgment is a weak fit, but an agent that holds clarification exchanges with customers on WhatsApp or email does interact directly with the people who write, and those people must be informed they are dealing with an AI system unless that is obvious from the context.

Contract review and sourcing are not among the Annex III high risk uses, so an internal assistant that makes no decisions about natural persons is minimal risk (with the Article 4 AI literacy duty). If a negotiation bot chats directly with supplier staff, Article 50(1) applies and it must tell them they are dealing with an AI system, unless that is obvious from the context. Public authorities using AI in procurement should still check national public procurement rules on transparency and equal treatment of bidders.

The classification and declaration work in the back office is minimal risk: classifying goods and preparing customs declarations is not listed in Annex III, which covers border control only where AI assesses natural persons (point 7). If the design adds a shipper facing assistant that asks for missing information in chat or by email, that assistant is limited risk and carries the transparency duty of Article 50 (people must know they are dealing with AI). The obligations that matter most come from customs law: the declarant stays responsible for the accuracy of the declaration whatever tool prepared it.

As a pure assist tool for agents it is minimal risk; the customer does not interact with the AI. It becomes high risk under Annex III point 4(b) if its data is used to monitor and evaluate individual agents' performance, and inferring agents' emotions at work is prohibited under Article 5(1)(f).

A shopping assistant interacts directly with people, so under Article 50(1) shoppers must be informed that they are dealing with an AI system unless that is obvious. It is not listed in Annex III, so it is not high risk. Manipulative or deceptive techniques that materially distort a shopper's behavior and cause significant harm are prohibited under Article 5(1)(a), which matters for how persuasion and urgency are designed.

A customer facing service agent must disclose that the customer is interacting with AI (Article 50). It is not high risk: it does not decide on access to essential services, credit or employment.

Article 50(2) puts a machine readable marking duty on the provider of the generative tool itself (Google, for example, marks Performance Max images with SynthID), with an exception only where the tool performs assistive standard editing or does not substantially alter the input. Article 50(4) separately requires the deploying brand to disclose when it generates or manipulates image, audio or video content that is a deep fake: content that resembles a real person, place or event and would falsely appear authentic, such as the synthetic "digital twin" likenesses used in the Lysol pilot. Content that is evidently artistic, creative, satirical or fictional is not exempt from that disclosure; the article only lets the brand make it in a manner that does not hamper the display or enjoyment of the work. The separate exception for content under human review or editorial control, with someone holding editorial responsibility, applies only to AI generated text published to inform the public on matters of public interest, not to image or video ad creative, so it does not exempt anything on this page. Lysol's pilot is a US deployment and the source does not describe it as operating under the EU AI Act.

A customer facing agent must make clear that people are dealing with AI, unless that is obvious from the context (Article 50). Initiating payments within a customer's mandate is not listed in Annex III. It becomes high risk if the same agent evaluates creditworthiness, for example by deciding on a buy now pay later or credit line at checkout (Annex III point 5(b)).

An AI system that interacts directly with people must be designed so that they know they are dealing with AI, unless that is obvious from the context (Article 50(1)). It is not high risk under Annex III as long as it does not evaluate eligibility for essential public assistance benefits and services (point 5(a)), creditworthiness (point 5(b)), risk and pricing for life and health insurance (point 5(c)) or emergency calls (point 5(d)). This holds only if emotion or vulnerability signals are inferred from what the customer says (text or transcript content), not from voice or other biometric features; an agent that infers emotion from a caller's voice is an emotion recognition system (Article 3(39)), which is high risk under Annex III point 1(c) and triggers the deployer disclosure duty in Article 50(3).

Article 50(1) requires an assistant that talks with people to make clear they are interacting with AI, unless that is obvious from the context. It is not listed in Annex III. The agent does allocate work, but it routes tickets to resolver and assignment groups based on the content of the request, not to individual workers based on their behavior or personal traits or characteristics, so Annex III point 4(b) does not apply. It also does not decide on recruitment, promotion, credit or access to essential services. Any use that assigns work to individual analysts, or monitors and evaluates them from their behavior or performance (including through the agent's logs), would need its own assessment.

Article 50(1): people must be told they are interacting with an AI system unless that is obvious. Finding locations and booking appointments does not fall under any Annex III category. If a healthcare version starts to triage patients by urgency, or a public body uses it to decide eligibility for a public service, reassess it against Annex III point 5.

Forecasting a company's cash flows is not listed in Annex III and makes no decision about a natural person, so the forecasting model itself carries no obligations beyond AI literacy (Article 4). The conversational layer interacts directly with treasury staff, so under Article 50(1) they must be informed that they are dealing with an AI system unless that is obvious from the context. Without a conversational layer the use case is minimal risk.

Article 50(1) requires providers to design AI systems that interact directly with people so that those people are informed they are dealing with AI, unless this is obvious from the context, as it usually is for an internal assistant. An analytics assistant that makes no decisions about people is not a prohibited practice under Article 5 and is not listed in Annex III. It would be high risk only if it were intended for an Annex III purpose, such as assessing the creditworthiness of natural persons (point 5(b)).

Setting or recommending retail prices for goods is not listed in Annex III and does not evaluate the creditworthiness, employment, or access to an essential service of a natural person, so this is minimal risk under the EU AI Act. It would need reassessment if a retailer used the same engine to set an individual price per identified customer rather than per product and store, which raises separate consumer protection and non discrimination questions the deployments on this page do not describe.

Forecasting product demand and ordering stock is not listed in Annex III. It would become high risk under Annex III point 4(b) only if the same system allocated tasks to employees based on their individual behavior or personal traits, or monitored and evaluated their performance, for example scheduling store staff by individual productivity. GDPR applies only when loyalty or customer level data feeds the forecasts; item and store aggregates on their own are not personal data.

Deciding whether to accept a commercial order is not listed in Annex III: it is not a creditworthiness, employment, essential service or biometric decision about a natural person. It stays minimal risk provided the decision is limited to a commercial transaction and does not extend into scoring the buyer's general creditworthiness or blocking access to an essential service.

The system analyzes shelves and products, not natural persons, and makes no decision about a person's access to a service, a job, credit or benefits, so it falls outside Annex III. If camera placement or software is changed to identify or track individual shoppers or staff, the analysis changes and biometric identification rules would apply.

Dispute processing between issuers, acquirers and merchants is not listed in Annex III and is not an evaluation of creditworthiness or credit scoring under Annex III point 5(b), and because cardholders do not interact with the system directly, the Article 50(1) transparency duty for AI that talks to people does not apply. Article 50(2) marking of generated text is a duty of the provider of the AI system that generates it, which includes an institution that builds its own dispute drafting agent and puts it into service under its own name. A drafted rebuttal built from attached case evidence performs an assistive function for standard editing of that evidence and does not substantially alter the underlying input, so it falls under the Article 50(2) exception and does not need machine readable marking. With that point checked, the tier stays minimal. A customer facing intake agent is assessed separately.

An internal data engineering tool that flags anomalies in pipelines and tables; it is not a use listed in Annex III and makes no decision about a natural person. If the monitored data feeds a high risk system, such as a credit or employment decision, the AI Act obligations attach to that downstream system, not to this monitoring layer.

Processing supplier invoices is not an Annex III use case, is not a practice prohibited by Article 5 and does not involve decisions about natural persons, so it is minimal risk and the AI literacy duty of Article 4 applies. Approvers who ask questions in chat use an internal tool they know is AI; if that is not obvious to the people using it, the provider must also inform them that they are interacting with an AI system (Article 50(1)).

Summarizing public threat reporting for an organization's own cyber defense is not an Annex III use, and recital 55 of the AI Act says components used solely for cybersecurity purposes do not qualify as safety components. The tier stays minimal while the assistant works on public sources and an analyst reviews and takes responsibility for what is published. It changes if the same tooling is used to profile or score employees.

Drafting business quotes for a seller to review is not an Annex III use and does not interact with the customer as an AI system. It would need a fresh assessment if the system set individual consumer prices or terms in areas such as credit or insurance, where Annex III point 5 can apply.

Matching a company's own incoming payments to its own open invoices is a back office finance operation. It is not listed in Annex III and does not decide a natural person's creditworthiness or eligibility for a service, so it is minimal risk and the AI literacy duty of Article 4 applies. Using deduction or payment behavior to score an individual sole trader's creditworthiness would need a fresh risk assessment.

Inline translation and rule based content targeting for internal communication do not decide on hiring, pay, promotion or termination and are not listed in Annex III, so they carry no specific obligation under the EU AI Act beyond general AI literacy, unless a deployer adds a feature that scores or ranks individual workers, which would then need separate assessment. A Blits.ai agent built on top of this platform is a conversational AI system, so it carries the Article 50 transparency duty to disclose that employees are interacting with an AI system, regardless of tier.

The system optimizes promotion terms between a manufacturer and a retailer; it does not decide a natural person's access to credit, employment, essential services or a regulated product, so it falls outside Annex III. It would need reassessment if a manufacturer used the same technique to set individually targeted prices or offers for identified consumers.