PrymeIntelligence

AI literacy and certification·Regulation

EU AI Act Article 4 AI literacy for AI agents in higher education

What Article 4 of the EU AI Act means for universities using AI agents: literacy for academics, admissions, student services and teaching assistants.

Universities using AI agents must, under Article 4 of the EU AI Act, take measures so that staff and others acting for them have sufficient AI literacy. In higher education that reaches academics, admissions teams, student services, and the doctoral students and teaching assistants who mark and tutor on the institution's behalf. The students affected shape what "sufficient" means.

What does Article 4 ask of a university?

Article 4 applies to every provider and deployer 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. Sufficiency is judged against 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 understand their opportunities, risks and possible harms. It is a duty on the institution as deployer, not on individual academics, though academics carry it out. Whether formal training is required is covered in does the EU AI Act require AI literacy training.

Several education uses are high-risk under the Act: deciding admission, evaluating learning outcomes, assessing the level of education a person should receive, and monitoring students during tests. Those bring Articles 14 and 26, discussed in Article 26 duties in higher education.

Who does it cover in a university?

  • Academic staff using agents to draft feedback, mark or pre-mark work, or design assessments.
  • Doctoral students and teaching assistants who mark or tutor on the institution's behalf and may use its agents to do so.
  • Admissions officers working with agents that sort or rank applications.
  • Student services staff using agents to answer queries about finance, accommodation or wellbeing.
  • Exam and academic integrity staff overseeing monitoring agents.
  • Professional services and IT teams connecting agents to student records.
  • Senior leaders and governing bodies accountable for academic standards.

Teaching assistants are the group most easily missed. They are often students themselves, on short contracts, and may mark hundreds of scripts with agent help. Who needs literacy under the Act in general is covered in who needs AI literacy under the EU AI Act.

What does literacy mean when a university agent acts?

University agents act in ways that change students' records and prospects. A governed marking agent might read submissions and rubrics at depth 1 and draft marks and feedback at depth 2, with every mark held for a named marker. A student services agent might answer routine questions at depth 3 and have anything touching fees, extensions or welfare held for staff.

Literacy for each group follows from that.

Markers and academics need to treat an agent's mark as a draft and own the final mark. They should know the agent's typical weaknesses, such as rewarding length or confident style, or missing an original argument that departs from the rubric. They should be able to explain a mark to a student without referring to the agent.

Admissions staff need to recognise when an agent's ranking leans on factors that stand in for background, such as school or postcode, and to read the provenance of an applicant's information: a grade confirmed by an examining body is signed, a personal statement is declared, an emailed reference is external until verified.

Student services staff need to recognise when a routine query is not routine. A question about deferring fees may signal financial hardship or a welfare concern. The agent should hold such cases for a person, and staff should know why a case was held.

Integrity staff need to know that a monitoring agent's flag is not evidence of misconduct.

Everyone needs to understand that an agent's certificate lapses after 90 days, and that a run that clears no depth drops it to depth 1 the same day, which during exam season turns automated work into manual work at the busiest time.

How does a university meet Article 4?

Requirement Control Evidence it leaves
Sufficient literacy for staff Core module plus role modules for marking, admissions, student services and integrity Records by role and date
Persons acting on its behalf Teaching assistants trained before marking with agents Training records linked to marking allocations
Fitted to the students affected Modules on bias, welfare signals and explaining decisions Assessment results
Awareness of risks and harms Training on typical marking and ranking errors Moderation results comparing agent and human marks
Kept current Briefings before each admissions cycle and exam period Briefing records
Leadership literacy Governing body session on where agents act Minutes

The governance angle, including why literacy has reached governing bodies, is covered in the EU AI Act made AI literacy a board question.

What will a regulator or quality body ask?

Market surveillance authorities, data protection authorities and quality bodies may ask: how do you decide what literacy each role needs? How are teaching assistants trained before marking with agents? Show moderation results comparing agent-drafted and final marks. How do student services staff recognise welfare signals in agent-handled queries? How is the governing body informed?

What does Article 4 not cover?

It does not set academic standards or assessment regulations. It does not replace Articles 14 and 26 for high-risk education uses. It does not prescribe course content or credentials. It does not govern students' own use of AI in their work, which is a matter for academic integrity policy. Some institutions use Pryme Intelligence Academy's Foundations badge as a shared baseline, since that track is open to students as well as staff, but measures must still fit the agents in use. A neighbouring sector comparison is in Article 4 in telecoms, and the general model in AI literacy and certification for people who work with agents.

This is an explanation of the rules, not legal advice; confirm how they apply to you with counsel.

Questions people ask

Are teaching assistants covered by the university's duty?

If they use the university's AI systems on its behalf, such as to mark, the university must take measures for their literacy, typically training before marking allocations.

Does Article 4 cover agents that answer student questions?

Yes. It covers any AI system the university deploys. Staff overseeing those agents need to recognise when a question is really a welfare or finance concern.

Should academics be trained on every agent?

On the agents they use. An academic who never uses the marking agent needs only baseline literacy; one who marks with it needs specific training.

Train the people who run your agents.

Pryme Intelligence Academy: free courses on the real product, audio lessons, labs in a free workspace, and certificates anyone can check on VerifyDoc.ai. Teams get seats, an admin view and a free readiness assessment.