Consumer lenders rely on brokers, intermediaries and collection partners as much as on their own staff, and Article 4 of the EU AI Act reaches all of them. Lenders using AI agents must take measures so that everyone using those agents on their behalf has sufficient AI literacy, fitted to their role and to borrowers who are often financially stretched.
What does Article 4 ask of a lender?
Article 4 applies to all providers and deployers of AI systems. In substance, they must take measures to ensure, to their best extent, a sufficient level of AI literacy among their staff and other persons dealing with the operation and use of AI systems on their behalf, taking into account those people's knowledge, experience, education and training, the context of use, and the people the systems are used on.
AI literacy means the skills, knowledge and understanding to use AI systems in an informed way and to be aware of their opportunities, risks and possible harms. In consumer lending the "people the systems are used on" clause matters: borrowers include people in financial difficulty, and literacy must include recognising when an agent's output would harm them.
Article 4 sits alongside heavier duties. Creditworthiness assessment of individuals is a high-risk use, so lenders also face Article 14 human oversight and Article 26 deployer duties for that part of the work. Whether Article 4 itself requires training is covered in does the EU AI Act require AI literacy training.
Who does it cover in consumer lending?
"Other persons acting on its behalf" stretches further in lending than in most sectors:
- Underwriters and credit reviewers deciding applications an agent assessed.
- Customer service staff answering borrowers with agent help.
- Collections staff contacting borrowers in arrears, often with agent-drafted messages.
- Brokers and credit intermediaries who use the lender's agents to submit or pre-assess applications.
- Outsourced collection agencies using the lender's agents or systems.
- Product, risk and data staff who set the agent's criteria and bounds.
- Senior managers and directors accountable for lending decisions.
The general mapping is in who needs AI literacy under the EU AI Act.
What does literacy mean when a lending agent acts?
A governed lending agent might read bureau data and bank transactions at depth 1, draft an assessment and a decision letter at depth 2, approve routine applications within tight criteria at depth 3, and have declines and edge cases held for an underwriter. In collections it might draft reminders at depth 2 and send templated ones within rules at depth 3, with fees and referrals held.
Literacy for lending staff therefore means:
- Reading an assessment, not a score. Knowing which facts drove the outcome and how strong each is. Income verified from bank data is sealed or signed; income stated on an application is declared. An assessment resting on declared figures deserves more care.
- Recognising vulnerability. Spotting signals of financial difficulty in agent-handled conversations, and knowing the agent should hold or refuse further automated contact when they appear.
- Knowing what the agent may not do. A refusal means the agent was asked to step outside its bounds, such as contacting a customer outside permitted hours. Staff should route refusals, not rephrase around them.
- Expecting change. An agent's certificate lapses after 90 days, and a run that clears no depth drops it to depth 1 the same day. When that happens, automated approvals and reminders become manual work.
For brokers, literacy is narrower but real: understanding what the lender's agent does with the data they submit, and that a declared figure they type in will be treated as declared.
How does a lender meet Article 4?
| Requirement | Control | Evidence it leaves |
|---|---|---|
| Sufficient literacy for staff | Core module plus role modules for underwriting, service and collections | Completion and assessment records by role |
| Persons acting on the lender's behalf | Broker and agency agreements require training before agent access | Agreement clauses; partner training evidence |
| Fitted to the people affected | Vulnerability recognition included for all borrower-facing roles | Assessment results on vulnerability cases |
| Awareness of risks and harms | Training on automation bias, declared data and proxy factors | Assessment records |
| Kept current | Briefing at each agent recertification or criteria change | Briefing records linked to agent changes |
A step-by-step approach to building this is in how to set up an AI literacy programme under the EU AI Act.
What will a supervisor ask?
Consumer credit supervisors and market surveillance authorities may ask: how do you train brokers and collection agencies who use your agents? Show the training for underwriters who decided a sample of declines. How do borrower-facing staff recognise vulnerability in agent-handled cases? What happens when the agent's criteria change? For the creditworthiness work, they will also test Article 14 and Article 26 competence.
What does Article 4 not cover?
It does not set affordability or responsible lending rules. It does not replace Articles 14 and 26 for the high-risk credit work. It does not prescribe course content or credentials. It does not cover borrowers' own literacy, though explaining decisions clearly to them is good practice and often required by other rules. Sector comparisons are in Article 4 in insurance and in payments.
This is an explanation of the rules, not legal advice; confirm how they apply to you with counsel.
Where the Academy fits
For teams of up to 25 people, Pryme Intelligence Academy offers a free readiness assessment: the team takes the Foundations test set in its industry, and the report shows scores by concept, the gaps, and the first agent to start with. For a lender, running it on a collections or underwriting team shows where literacy is thin before agents go live. Our guide to AI literacy and certification for people who work with agents explains the levels.
Questions people ask
Do independent brokers count as acting on the lender's behalf?
If they use the lender's AI systems in the lender's process, they are dealing with its AI on its behalf in a practical sense. Lenders should set training expectations in broker agreements and check them.
Does Article 4 apply to collections agents?
Yes, if the lender or its agency uses AI systems in collections. Literacy for collections staff should weight vulnerability recognition heavily.
What is sufficient literacy for an underwriter?
Enough to read the agent's assessment, judge the strength of its evidence, recognise its weak segments and decide against it when right. For credit decisions, Articles 14 and 26 raise the bar to demonstrated competence.