Framework for the Use of AI in Language Education

Principle I

Safe use of AI in language education

Are our uses of artificial intelligence in language education legally compliant, transparent, and risk-free?

Guideline 1

Comply with existing legal requirements and institutional standards

Exercise caution with terms of service, age restrictions (typically 13+ or 18+) and institutional procurement policies
Ensure learners are legally competent to consent; seek parental permission where required, including language-accessible terms
Attribute AI-generated content appropriately and protect learners' intellectual property
Teachers retain responsibility for outcomes when delegating tasks to AI; maintain accountability and transparency (FATE)
EU AI Act 2025 prohibits emotion recognition, real-time facial recognition, and behaviour manipulation in schools; high-stakes AI applications require strong human oversight

Guideline 2

Safeguard the data security and digital safety of all participants

Understand what data (voice recordings, writing samples, metadata) AI tools collect, and vet tools for GDPR / FERPA compliance
Minimise personal data in prompts; avoid sharing identifiable learner work; understand data retention policies
Be alert to data mining risks: user content may be repurposed for model training, profiling or commercial use
Protect vulnerable learners — underage, migrants, refugees — from data misuse; limit AI use to trusted applications
Raise awareness among learners, colleagues and parents; advocate for institutional data-protection policies

Key competences (1–6)

Legal alignment IP adherence Teacher accountability Data identification Privacy awareness Protecting vulnerable learners

Did you know?

The 2025 EU AI Act prohibits emotion recognition and real-time facial recognition for monitoring in schools (Article 5). High-stakes uses — automated grading, admissions — require strong human oversight, explainability, and clear appeals processes. AI agents must be disclosed as non-human (Article 50).

Principle II

Responsible use of AI in language education

Does our use of AI position our teaching as a force for positive change?

Guideline 3

Promote social justice and linguistic and cultural diversity

Critically evaluate AI outputs for bias, stereotypes, cultural hegemony and epistemic injustice embedded in training data
Deliberately prompt AI to surface marginalised languages, varieties and perspectives; treat linguistic bias as a pedagogical opportunity
Affirm learners' cultural and linguistic repertoires; resist reduction to instrumental goals that marginalise identity
Address inequalities in infrastructure access (electricity, internet) and develop TPACK to mitigate digital divides
Move beyond critique to transformation — use AI as a lever for equity and inclusion rather than mere caution

Guideline 4

Foster environmental sustainability

Be aware of AI's carbon, water and energy footprint — infrastructure, model training and hardware production all carry significant environmental costs
Prefer calibrated, purposeful AI use over routine default use; evaluate when lower-impact alternatives suffice
Use existing open image banks or content repositories rather than generating new AI images where possible
For information-seeking tasks, conventional web searches use 10–30× less energy than AI-generated content
Encourage AI vendors to disclose environmental metrics; advocate for institutional procurement accountability

Key competences (7–12)

Sustain linguistic diversity Challenge bias in AI output Equitable access Environmental awareness Minimise footprint Model sustainable practice

Did you know?

Even advanced LLMs designed to reduce bias still exhibit implicit biases — associating negative language with certain racial groups, linking women more often with the humanities over science, and favouring men in leadership contexts (Maslej et al., 2025). Generating a single AI image uses roughly as much energy as charging a smartphone.

Principle III

Purposeful use of AI in language education

How does AI add genuine pedagogical value and improve the quality of teaching and learning?

Guideline 5

Enhance theory- and policy-informed language teaching pedagogy

Prioritise quality of AI use over quantity of tools; AI-assisted activities should foreground meaning-making and authentic language use
Use conversational agents for low-stakes communicative practice, particularly where target-language interaction is limited
Leverage AI for differentiated instruction: personalise goals, content and tasks to learners' levels, interests and needs
Harness speech-to-text and text-to-speech functions for inclusive multimodal learning and accessibility
Redesign assessment: shift focus from knowledge artefacts toward the learning process, communicative competence, and AI-transparent feedback

Guideline 6

Create expanded language learning opportunities

Ensure AI use genuinely transforms (not merely substitutes) existing activities — apply the SAMR framework as a diagnostic
Expand learner opportunity: enable communicative practice in languages with scarce human interlocutors; provide personalised feedback at scale
Enhance learner agency: support self-directed goal-setting, self-evaluation and meaningful choice over learning pathways
Challenge classroom monolingualism via AI-supported translation, plurilingual scaffolding and cross-linguistic comparison
Expand teacher capacity: use AI to notice what would otherwise be missed and to explore pedagogical approaches previously out of reach

IDEA Prompt Framework (Park & Choo, 2024)

Include essential PARTS (Person · Aim · Recipient · Theme · Structure) · Develop with CLEAR prompts (Concise · Logical · Explicit · Adaptive · Restrictive) · Evaluate and REFINE iteratively · Apply with accountability

Key competences (13–19)

Authentic language use Clear learning goals Personalised learning AI-transparent feedback Transformative tasks Learner agency Teacher capacity

Did you know?

The SAMR framework (Puentedura, 2006) distinguishes Substitution (no functional change), Augmentation (functional improvement), Modification (substantial redesign), and Redefinition (new tasks inconceivable without AI). Purposeful use targets Modification and Redefinition.

Principle IV

Human-centred use of AI in language education

How can teachers and learners use AI reflectively and maintain professional agency?

Guideline 7

Prepare learners for critical, healthy engagement with AI

Develop AI literacy through existing disciplinary strengths: critical text evaluation, register, accuracy and ideological analysis transfer directly to AI output
Compare human- and AI-generated texts; extend collaborative critique of hallucinations and embedded bias as language learning activity
Monitor learner well-being: watch for fatigue, disengagement and over-reliance associated with excessive AI use; calibrate screen time and cognitive load
Affirm learner identity, voice and plurilingual repertoire — frame AI as a foil or drafting resource, not a model or authority
Leverage AI's multilingual capabilities for dynamic cross-linguistic practice, code-switching and meaning-making across languages

Guideline 8

Empower language teachers as reflective professionals

Use AI for reflection-for-action: generate lesson variants with different degrees of autonomy; explore counterfactual scenarios
Support reflection-in-action: deploy AI tools to monitor AI-assisted activities and make real-time pedagogical judgements
Use conversational agents and transcript analytics for reflection-on-action after lessons or instructional sequences
Leverage AI to support classroom-based inquiry — action research, exploratory practice, teacher-led scholarship — through collaborative refinement of research questions and systematic transcript analysis
Cultivate reflexivity: use AI to make the values and assumptions embedded in pedagogical choices visible, deliberate and professionally articulable

Example reflective prompt (Guideline 8)

"Assume the role of a critical friend and experienced language educator. Your task is to help me reflect on a lesson I taught today … Ask probing questions about: (a) what worked well and why, (b) what to do differently in the future, and (c) how AI influenced the dynamics of the lesson."

Key competences (20–25)

AI literacy integration Learner well-being Affirm learner identity Reflection-for-action Classroom inquiry Critical reflexivity

Did you know?

Drawing on Schön (1983) and Farrell (2022), the framework distinguishes three modes of reflective practice: reflection-for-action (planning), reflection-in-action (real-time judgment), and reflection-on-action (retrospective analysis). AI can support all three.