Career and Technical Education AI Playbook

Built on Mississippi’s Statewide AI Framework

The Mississippi Statewide AI Framework (August 2026) is the foundation for this playbook. It supplies shared priorities and a learning progression; this playbook applies them to career and technical instruction, workforce training and placements. Both are voluntary. Applicable law, adopted standards and institutional authority remain controlling.

Read the August 2026 Statewide AI Framework (PDF)

Learning progression. High school, community college/technical programs and workforce entry points. Apply the relevant course or credential conditions; pathways need not form one sequence.

Five priorities in practice

Provide an equivalent, accessible activity (1); assess integrity and verified performance (2); protect student and employer information (3); connect practice to occupational work (4); review demonstrated skill and the full instructional burden (5).

Priority numbers refer to literacy and access (1), ethics and critical thinking (2), privacy and governance (3), workforce readiness (4), and strategic alignment and measurable outcomes (5).

Selected Framework domains, practice and evidence

Framework domain	Practice and evidence to retain
10. Sector and Pathway Awareness	Map one activity to its occupational pathway and current curriculum; retain Tool 1.
3. Critical Evaluation of AI Outputs	Recalculate, check procedures and reject unsupported claims; retain Tool 3.
8. Human Agency, Judgment, and Restraint	Demonstrate the learner's skill under authorized conditions; retain performance evidence.
2. Data Literacy and Data Stewardship	Agree on workplace inputs and evidence handoff; retain the placement agreement.

Keep the shared human-review boundary

Apply the Framework's human-review triggers (p. 35) before action: assessment or credentials; financial allocation or compensation; health or physical safety; discipline or employment; legal or compliance content; and sensitive or protected information. Its additional agent triggers cover autonomous system-of-record changes; persistent memory; tool access, credentials or external transactions; handoffs in safety-critical, regulated or student-facing processes; missing stop conditions or logged escalation to a named human owner; and institutional responses to synthetic media about identifiable people. Product or pilot approval does not replace action-specific review. Review the system and access before enabling them; review each proposed system-of-record change before execution. Advance permission cannot waive these triggers.

Use evidence within its limits

Use the crosswalk to select practice and retain observable evidence. It is MAIN's implementation mapping, not a validated proficiency test, credential or certification. Local rubrics and pilot measures alone do not establish statewide proficiency or causal impact. Recheck alignment when the Framework changes; record the edition, local owner and next review.

Source: August 2026 Framework, pp. 5-7 (priorities), 14-30 (domains and progression), and 35-37 (review, measurement and maintenance).

Browse all MAIN AI playbooks · Published October 15, 2026. Evidence cutoff: October 13, 2026.

Cover illustration: AI-generated editorial artwork, not a documentary image. Conceptual technical-training workbench with an unpowered practice board, caliper, workbook and laptop; not an actual institution.

Download the CTE Playbook (PDF · 21 pages)

On a small screen, wide tables scroll horizontally. Focus a table to use the arrow keys.

In this playbook

Mississippi Artificial Intelligence Network (MAIN)

Start with the skill students must demonstrate

AI belongs in career and technical education when it helps students practice, inspect, explain, or improve work they can take responsibility for. A convincing explanation, polished report, or plausible troubleshooting suggestion does not establish occupational competence. Build each activity around a visible performance and a source the student can check.

Use this playbook when a secondary CTE instructor, community college faculty member, workforce trainer, program administrator, or work-based learning coordinator wants to introduce AI into a course or placement. It produces a bounded instructional pilot: a competency map, an assignment with explicit AI permissions, an assessment plan, a workplace agreement where needed, and a decision about continuing the activity.

For a first planning session, bring one existing assignment, the curriculum or training standard governing it, the current scoring criteria, and the institution’s technology policy. The instructor and program lead can begin the curriculum map together; involve IT, privacy, accessibility, and a workplace supervisor before decisions requiring their authority. A suggested 45–60 minutes is a planning allowance, not an implementation deadline. Leave with one activity selected and unresolved approvals assigned to named owners.

This playbook supplies MAIN’s recommended instructional method. It does not authorize a product, change a credential requirement, approve disclosure of student or employer information, or replace a qualified instructor, supervisor, manufacturer procedure, or safety program. The examples are hypothetical teaching scenarios, not reports of Mississippi programs or measured results.

Choose your starting path

Start with Tools 1–4 for a classroom activity. Add Tool 5 only when a workplace is involved. Tools 6–7 govern the pilot decision and changes; instructors do not need to complete every record before discussing an idea.

1. Ground the activity in Mississippi’s actual curriculum

Secondary and postsecondary CTE have different curriculum authorities. Mississippi State University’s Research and Curriculum Unit maintains a secondary curriculum directory; the Mississippi Community College Board provides the postsecondary curriculum directory. Use the correct directory to locate the framework for the particular program, then have the program lead confirm the edition in force locally. A directory entry alone does not establish every course requirement or current certification rule. RCU secondary curriculum, MCCB curriculum.

The Governor’s May 21, 2026 announcement describes the Mississippi Statewide AI Framework as a guide spanning K–12, postsecondary education, and workforce learning, and explicitly distinguishes it from a curriculum or mandate. Use the August 2026 Statewide AI Framework and the crosswalk above as the AI-literacy foundation; retain the approved course framework as the source of course requirements. Official framework announcement.

For a secondary program, the program lead confirms district technology rules and applicable MDE guidance. A postsecondary program uses its college’s academic, technology, accessibility, and clinical or placement rules. In a short-term workforce course, the sponsor confirms the intended job task and employer conditions. A dual-enrollment activity needs agreement about which institution governs accounts, evidence, and assessment. Do not assume that one permission transfers across these settings.

Write the exact competency and its source before writing a prompt. Identify three kinds of evidence:

These are MAIN’s assessment categories, not a new state credential framework. Independent performance does not require withdrawing an accommodation or assistive technology. Determine permitted aids from the learning objective, accommodation plan, and applicable assessment rules.

Check external credential assessments separately from classroom assignments. Have the program lead obtain the credential issuer’s applicable rules on permitted aids and accommodations, record their source and date, and communicate the assessment conditions in advance. Permission to use AI for a class activity does not establish permission to use it on a certification examination.

For example, a business communication activity may allow a first draft from AI while requiring the student to verify every service claim. An equipment procedure activity may allow students to critique an AI explanation in class while retaining the approved procedure as the sole operational instruction. Different objectives require different permissions.

Teach AI literacy through the occupational task

Before the first activity, check whether students can distinguish a calculator or fixed rule from generated language, identify which facts the tool received, explain why a plausible answer can be wrong, and recognize information they may not disclose. Use a brief explanation or critique exercise to find gaps; do not infer readiness from frequent chatbot use.

Novices can compare an instructor-provided draft with a short source and identify one unsupported claim. More experienced learners can specify constraints, compare alternatives, document corrections, and explain when to stop. Advanced learners can analyze how an AI step fits a workflow and what evidence a supervisor would need. The occupational skill stays central at each level.

Tool 1: Competency and activity map

Use when: selecting or revising one instructional activity. Completed by: instructor and program lead. Bring: applicable framework, existing assignment, assessment rules, and approved-tool policy. Produces: a decision to proceed, revise, or defer. Complete when: the performance, permitted assistance, verification source, and accountable reviewer are explicit. Stop when: the curriculum edition, safety authority, assessment rules, or tool permission cannot be confirmed.

Record:

• Program/course and level: ______
• Competency, document title, edition, section/page, and location: ______
• Performance students must demonstrate: ______
• Permitted aids and accommodations: ______
• AI’s specific instructional role: ______
• What students must do independently: ______
• Evidence students submit: ______
• Authoritative source or observation used to verify: ______
• Instructor responsible for acceptance: ______
• Decision, unresolved issue, responsible owner, and review date: ______

Do not approve the activity because it is engaging alone. Proceed when its evidence can show the intended competency and the instructor can detect consequential errors.

2. Choose assistance the instructor can verify

Begin with a task whose output remains inside instruction and whose correctness can be checked before anyone acts on it. Useful starting points include alternative explanations of an assigned concept, practice interviews, critique of a fictional work product, or a communication draft based on approved source material. None is automatically suitable for every learner or tool.

Classify the proposed use by its consequences, not by whether it looks like “just text.” A generated calculation may enter a quote; a summary may omit a warning; a translated instruction may alter its meaning.

The last row is a boundary for this playbook’s pilot method. It is not a claim that one blanket law bans every AI use in those areas.

Do not connect an instructional assistant to machinery, vehicle controls, clinical systems, or production accounts as a shortcut to realism. Students can learn the limits of automation through simulated cases and controlled demonstrations.

3. Establish the tool and data boundary first

Obtain institutional approval for the actual product, account type, configuration, and learner age group. Ask who can access submitted material, whether it is retained or used for training, what administrators can control, and how accounts and records are closed. An instructor’s personal subscription is not evidence of institutional approval.

For student education records, the U.S. Department of Education advises checking institutional approval and consulting IT. Its guidance identifies conditions for using the FERPA school-official exception, including institutional control of record information and restrictions on unauthorized reuse or disclosure. Treat that exception as a determination for the institution, not an automatic property of an AI service. Removing a name alone does not establish that disclosure is safe or lawful. Department of Education guidance on classroom applications.

For the first pilot, use instructor-created fictional records or materials already approved for that use. Do not upload grades, individualized student records, real patient information, employer drawings, credentials, or proprietary process data merely to make an example more authentic. Confirm that permission to read a document also permits the proposed upload or reuse; those permissions may differ.

Check age restrictions and account arrangements before requiring students to register. Offer an equivalent route to the same learning objective when tool access is unavailable or unsuitable. Test keyboard navigation, assistive technology, readable output, and any required audio or visual interactions with the institution’s accessibility support. A text interface alone is not proof of accessibility.

If an activity includes children under 13, have the institution review COPPA applicability and consent arrangements with the provider. FTC guidance limits a school’s ability to consent on a parent’s behalf to the educational context, rather than unrelated commercial use. That review is separate from FERPA and from a product’s contractual age restrictions. Do not treat a parent’s permission or an instructor’s approval as automatic clearance for every service. FTC COPPA FAQ, school section.

Before choosing a product, have the program lead and IT review an actual sample task with the intended account. Record the evidence for data handling, age/account eligibility, administrative controls, accessible use, export/deletion, support, and total cost including verification. If a necessary condition remains unknown, use instructor-provided output or defer student account use. A “free” account may still add cost through devices, supervision, support, or required subscriptions.

Make the product decision inspectable

Use this check before requiring an account or changing a classroom tool. The program lead coordinates IT, privacy, purchasing, and accessibility owners as applicable. Bring the proposed account/configuration, institutional policies, actual supplier terms, a sample assignment, and the equivalent activity. Record evidence and a decision for each row; a product description alone does not complete the check.

The check produces a scope-specific proceed, revise, or defer decision. Record the tool/account, permitted learners and inputs, activity, decision owner/date, evidence locations, exclusions, and review trigger. Complete it when every necessary condition has evidence and its authorized owner has resolved it. An unresolved necessary condition stops the affected use; it does not require stopping a suitable offline lesson.

Tool 2: Assignment permission card

Use when: issuing an AI-related assignment. Completed by: instructor after required institutional review. Bring: Tool 1, approved account configuration, and relevant policies. Produces: a student-facing instruction that can be used without guessing. Complete when: students know what they may do, what evidence to submit, and how to use the equivalent route. Stop when: required access, data permissions, accommodations, or instructor review are unresolved.

Record and give to students:

• Task and learning objective: ______
• Approved tool/account, or instructor-provided output: ______
• AI may assist with these steps: ______
• These steps must be completed under the stated independent-performance conditions: ______
• Permitted inputs and prohibited inputs: ______
• Required verification sources: ______
• Required disclosure of assistance and retained evidence: ______
• Equivalent activity/accessibility arrangements: ______
• Assessment criteria and any safety-critical criteria: ______
• Where to ask for help or report an unsafe output: ______

Make permissions specific. “AI is allowed” leaves students uncertain whether they may generate a whole answer, revise a sentence, use a translation, or request feedback.

4. Teach a repeatable verification habit

Use the same sequence across assignments so students practice judgment rather than collecting prompt tricks:

1. State the task. Identify the work product, audience, and constraints. Separate the facts supplied by the instructor from assumptions.
2. Produce a bounded draft. Request a specific explanation, comparison, or draft using approved inputs. Ask the tool to identify missing information, but expect that it may fail to do so.
3. Check against evidence. Open the actual source. Recalculate quantities. Compare a procedure with the approved procedure. Test code in an isolated environment. A citation generated by the tool is a lead to investigate.
4. Decide and correct. Keep, revise, reject, or escalate each consequential claim. Record why. If a critical claim cannot be verified, stop its use.
5. Demonstrate the skill. Explain the final decision or perform the targeted task under the assignment’s conditions. Submit enough evidence for the instructor to inspect the judgment.

For software exercises, use an isolated environment, review the code, and test its behavior. These instructional safeguards adapt NIST SP 800-218, SSDF v1.1, practices PO.5, PW.7 and PW.8.

A second model’s agreement is not independent confirmation. A confident answer or self-reported confidence percentage is not a safety determination.

Hypothetical instructional prompt: “Using only the instructor-approved excerpt below, draft a plain-language explanation of the concept for a first-year learner. Separate statements supported by the excerpt from questions needing the instructor. Do not supply equipment operating steps, missing specifications, or invented references.”

This prompt communicates the task. It does not guarantee compliance. The instructor still reviews the output before using it with students.

Tool 3: Student verification record

Use when: submitting AI-assisted work. Completed by: student; checked by instructor. Bring: original draft, approved sources, and assignment criteria. Produces: an inspectable record of judgment. Complete when: each consequential claim has evidence and a disposition. Stop when: a safety-critical claim cannot be verified, a prohibited input was disclosed, or an unsupported instruction would affect real work; pause the affected step and involve the instructor.

Also record the tool and date used, permitted assistance received, unresolved questions, and final work location. Retain selected excerpts or a redacted record under institutional rules; do not require unrestricted public sharing of conversations.

5. Assess competence and assistance separately

An AI-assisted product can look strong while hiding weak understanding. Conversely, a student can demonstrate sound technical judgment while writing less polished prose. Score the performance the course intends to teach.

Use direct observation, explanation, a demonstration, or a comparable task to confirm the learner’s competence. Tell students the conditions in advance. Do not turn the confirmation into a surprise withdrawal of approved aids.

Tool 4: Evidence-based assessment guide

Use when: scoring a pilot assignment. Completed by: instructor. Bring: Tool 1, Tool 2, student work, verification record, and performance evidence. Produces: competency feedback and a documented grading decision. Complete when: the decision is supported by observable evidence. Stop when: safety-critical performance is unacceptable or evidence is insufficient; use the course’s established remediation and grading process.

Set course-specific weights and pass criteria before instruction. Treat safety-critical criteria as separate acceptance conditions where the course requires them; a high writing score cannot offset an unsafe performance. Do not use an AI detector score as the sole basis for a misconduct finding. Review the actual work, student explanation, evidence, and institutional process.

6. Apply the method in six CTE settings

Each scenario below is hypothetical. These are suggested instructional designs, not adopted curriculum requirements or proven outcomes.

Worked lesson: check a supplier estimate before communicating it

Hypothetical lesson with synthetic teaching data. This is a business-math and workplace-communication exercise. The values are deliberately created for instruction; they are not a real quote, supplier price, or purchasing recommendation. Before using it for a grade, the instructor maps it to the actual applicable course competency through Tool 1.

Learning objective: calculate a stated subtotal, identify missing terms, and write a source-supported request for clarification. The student demonstrates the calculation and explains the unresolved conditions.

Source supplied to students: a fictional supplier note says eight training kits cost $25 each and delivery costs $12. It says nothing about tax, availability, delivery date, or certification. The AI-style draft supplied by the instructor says: “The eight certified kits will arrive tomorrow for a final total of $212, including tax.” This is a deliberately flawed teaching draft, not output claimed to have come from a tested product.

Permission card, completed for this lesson: students may use an institution-approved tool to revise the wording of a clarification request. They may not invent terms, send a message, make an order, or upload real student/employer information. Students without an approved account receive the same draft on paper or through an accessible document. Both routes use the same source and assessment criteria.

Lesson sequence:

1. Students read the source and calculate the known charges with the aids authorized for the course: eight times $25 plus $12 equals $212.
2. They mark each consequential statement in the draft and complete the verification record below.
3. They write a clarification request, using approved assistance if available, and explain why a known subtotal is different from an authorized final price.
4. The instructor asks each learner to explain the calculation and one unsupported claim. If evidence of competence is insufficient, use the course’s normal feedback and reassessment process.

Acceptable revised message: “The stated kit and delivery charges total $212. Please confirm availability, delivery date, any tax or additional charges, and certification documentation if applicable before an order is considered.” No message is sent during the exercise.

Instructor acceptance criteria: the arithmetic and units are correct; every unsupported assurance is identified; the message preserves uncertainty; and the student can explain the decision under authorized assessment conditions. Polished wording cannot compensate for an unsupported assurance. These are lesson criteria, not a statewide rubric or credential rule.

What to retain: calculation, selected verification entries, final message, permitted assistance disclosure, and the instructor’s competency decision. Do not require every chat turn. If a learner misses the tax distinction, reteach the distinction and reassess; do not assume misconduct. Measure the activity’s value through learning evidence and total teaching effort, rather than assuming that using AI caused improvement.

Use the result for feedback: if the calculation, claim checks, message, and explanation meet the stated criteria, record that the lesson objective was demonstrated. If the learner gets the arithmetic right but preserves an invented assurance, return the message for correction and reassess the explanation. If the learner cannot explain the calculation under authorized conditions, record that the evidence is insufficient and use normal remediation. A corrected draft alone does not erase the need to demonstrate the targeted skill.

Advanced manufacturing: inspect a flawed explanation

Situation: students are learning to distinguish observations from troubleshooting hypotheses. Decision: use AI to supply a critique exercise, while keeping operational decisions with the instructor. The instructor prepares fictional symptoms and a deliberately incomplete AI explanation, then removes any actionable unsafe instructions before distribution.

Students identify what is known, what was assumed, which reference would be needed, and what question belongs with the qualified instructor. They compare the explanation with approved course material. The work product is a corrected evidence table and oral explanation, not a new machine procedure.

Verification and boundary: the instructor scores the reasoning and any separately authorized practical demonstration. No learner operates equipment on the basis of the generated explanation. OSHA’s hazardous-energy resources explain the importance of established procedures and training in covered workplaces; applicability and instructional lab rules require the appropriate local authority. This playbook does not prescribe lockout/tagout steps. OSHA hazardous-energy overview.

Automotive or diesel: separate diagnostic evidence from guesses

Situation: a fictional training case contains a symptom but lacks identifying equipment information. Decision: students may use AI to list missing information and possible explanations, then reject any generated specification that lacks the correct approved reference.

Students record the observations they would need and consult instructor-supplied service information. The instructor checks that the learner can explain why a generic specification cannot be substituted for one applicable to the actual equipment. The final work is an evidence-based diagnostic discussion. No generated torque value, wiring instruction, or repair authorization enters actual work.

Healthcare: practice communication with a fictional patient

Situation: students practice explaining an instructor-approved topic to a fictional patient. Decision: permit a language draft while retaining clinical content and acceptance with the instructor.

Students compare the draft with the supplied material, identify unsupported additions, and demonstrate a respectful explanation in a role-play. If a translation is used, a qualified reviewer checks its meaning before it becomes instructional guidance. The output is a reviewed communication exercise. Real patient records, diagnosis, treatment recommendations, and clinical decision-making are outside this activity.

Construction: audit a quantity estimate

Situation: an instructor supplies a fictional drawing and clearly labeled assumptions. Decision: students can critique an AI-generated material estimate rather than accept its arithmetic.

Students calculate quantities independently with authorized aids, check units and omissions, and reconcile differences. The instructor reviews the method and distinguishes a classroom estimate from a usable project quote. Missing drawing details remain questions. The exercise does not certify code compliance, structural adequacy, or suitability for procurement.

Culinary or hospitality: revise a communication without inventing assurances

Situation: students draft a response to a fictional guest question using an instructor-approved menu and policy. Decision: allow a communication draft but require checking every factual assurance.

Students remove claims the source cannot establish and identify when a supervisor must answer. The instructor checks the final message and the student’s reasoning. The exercise does not permit guessing ingredients, allergen absence, food temperatures, or accommodation availability. Those facts require the appropriate approved source and responsible person.

Information technology: explain and test a small program

Situation: students receive a small instructional program with an error. Decision: allow AI explanation and a proposed change inside an isolated environment with no real credentials or production connection.

Students inspect the change, run meaningful tests, explain its behavior, and complete a comparable skill demonstration. A passing sample test alone does not establish security or correctness. The instructor accepts the work against the course criteria. Installing packages, changing the environment, or executing generated commands requires the assigned permissions; the tool cannot grant itself those permissions.

Security basis: NIST SP 800-218, Secure Software Development Framework v1.1, practices PO.5, PW.7 and PW.8. MAIN applies these principles to a supervised instructional exercise; this example does not certify software security or authorize production access.

7. Agree on workplace use before placement

Ask the program advisory group what changes on the job

Before widening the pilot, ask employers and program advisory members to distinguish tools used in hiring from tools used in actual work, tasks a new worker must perform without generative assistance, and tasks where verified assistance is welcome. Ask for a nonconfidential example of acceptable work and the error that would make it unacceptable. Record disagreements and the date of the discussion; one employer’s preference is not a statewide occupational requirement.

Translate feedback into a task, assessment criterion, or permission change. A useful portfolio entry can show a permitted work sample, a correction, the source used, and the learner’s explanation of responsibility. It should not imply that prompt-writing alone qualifies a student for an occupation.

For regional coordination, AccelerateMS describes its ecosystem approach as connecting labor-market information, training resources, and local stakeholders. Use the current contact route to identify an appropriate workforce partner when the program lacks one; participation does not confer endorsement or replace institutional authority. AccelerateMS regional coordination.

An institution’s classroom approval does not establish an employer’s permission. Before a work-based learning assignment includes AI, the coordinator and supervisor should agree on the task, information boundary, permitted product/account, review responsibility, and student evidence. If the employer cannot authorize the proposed data use, redesign the assignment with fictional or approved information.

Keep workplace verification practical: the supervisor confirms the relevant performance, and the instructor assesses the learning evidence. Do not transfer confidential workplace records into a student portfolio to prove authenticity. A redacted description or supervised demonstration may provide adequate evidence when both parties agree.

Tool 5: Work-based learning AI agreement

Use when: an AI-assisted activity touches a placement. Completed by: coordinator, instructor, and authorized workplace supervisor. Bring: placement requirements, employer policy, Tool 1, and Tool 2. Produces: a permitted task and evidence handoff. Complete when: all parties understand their responsibilities and the student receives instructions. Stop when: employer permission, confidentiality, supervision, or safety requirements conflict.

Record:

• Task and competency: ______
• Approved tool/account and where use may occur: ______
• Permitted inputs; information that must remain at the workplace: ______
• Draft reviewer and person authorized to accept workplace output: ______
• Student evidence retained by the institution: ______
• Permitted independent and assisted performance conditions: ______
• Stop/report route for student and supervisor: ______
• Agreement owners, date, scope, and review trigger: ______

No student should have to choose between completing the assignment and violating an employer’s information rules. Provide a supervised or simulated alternative when needed.

8. Run a pilot that can support a decision

Select one assignment and a group small enough for the instructor to inspect the work. Establish the existing method’s performance first, using a comparable activity where feasible. Differences in task difficulty, learner experience, and instructor support can change results; record those differences rather than treating a simple before-and-after comparison as causal proof.

Measure the whole activity: preparation, learner effort, verification, rework, instructor grading, access support, and any workplace review. Report acceptable performances as a count out of all attempted performances. Record withdrawals, access failures, unresolved errors, and safety stops. A smaller drafting time alone does not establish instructional value.

Tool 6: Pilot charter and review decision

Use when: beginning and reviewing the activity. Completed by: program lead with instructor and relevant support owners. Bring: Tools 1–5 as applicable, baseline evidence, and local policies. Produces: a continue, revise, pause, or stop decision. Complete when: the evidence and decision owner are recorded. Stop when: a protected-data incident, unsafe recommendation used in practice, unmet accessibility need, or lost supervision makes the activity unacceptable.

Record:

• Assignment, participants, timeframe, and accountable owner: ______
• Baseline/comparison method and its limitations: ______
• Learning criteria and thresholds selected before the pilot: ______
• Workload, cost, access, error, and rework measures: ______
• Review capacity and who can halt the activity: ______
• Results with denominators and unresolved issues: ______
• Learner and instructor feedback, including the equivalent route: ______
• Decision and rationale; next review date/trigger: ______

Continue only when competence remains demonstrable, verification is workable, information use is permitted, and the institution can sustain review and access. Expand by adding one changed condition at a time, such as another assignment or learner group. Recheck the boundary before each expansion.

Include learners using the equivalent route in the review. Where devices, connectivity, scheduling, or paid access limit participation, use supervised in-class access, instructor-provided outputs, or an offline comparison exercise. Do not require learners to buy an account or assume statewide conditions from one class. Record the barrier and whether the alternative still produces the required competency evidence.

Before expansion, the program lead checks faculty preparation using one intentionally flawed output, a sample assessment, and the reporting route. Record who pays for accounts and support, who replaces an unavailable instructor, and what happens if the product changes. Any proposed grant-funded purchase needs the actual funding authority’s allowability and procurement review; this playbook makes no claim that AI subscriptions are automatically eligible under Perkins or another program.

9. Stop, correct, and maintain

Tell students how to stop using an output and contact the instructor. If material may have been disclosed improperly, follow the institution’s reporting process; students should not be asked to investigate the incident themselves. Correct affected instruction, identify work requiring recheck, and preserve relevant evidence according to institutional rules.

Tool 7: Change and correction record

Use when: an error, incident, or material change occurs. Completed by: program lead, with designated support owners. Bring: affected assignment, permitted evidence, and institutional response requirements. Produces: a correction and a decision on resuming use. Complete when: affected work is addressed, the revised boundary is tested, and the authorized owner records the disposition. Stop when: the information exposure, technical error, or safety consequence remains unresolved.

Record the trigger; affected users and work; immediate containment; reporting owner; corrected content or configuration; recheck evidence; people informed; and resume, revise, or retire decision. Review the activity when the curriculum, credential assessment, employer policy, tool terms, account configuration, accommodation needs, or verification source changes.

Source and method record

The linked agency and curriculum pages support the specific contextual and boundary statements beside them. The activity sequence, assessment categories, tools, scenarios, and pilot decisions are MAIN recommendations prepared for local adaptation. They have not been validated through a field trial, and this manuscript makes no claim of improved learning, employment, or productivity.

Software-security source: NIST SP 800-218, SSDF v1.1 (final, February 2022). PO.5 addresses protected development environments; PW.7 addresses code review; PW.8 addresses executable-code testing. This voluntary guidance is adapted here for instruction.

Evidence cutoff: October 13, 2026. Reconfirm the applicable curriculum, credential assessment rules, institutional approvals, and tool conditions before instruction.