› Grades 9–12 · ages 14–18 · Policy debate
Should AI development be regulated?
“Governments should require licenses and safety testing before frontier AI models are released.”
A policy debate on frontier-model regulation that takes both catastrophic and everyday risk seriously, and then asks whether a license regime would reduce either. Students encounter alignment, red-teaming, and regulatory capture, and must weigh a plan against the world without it.
60 minutes · one class period · free to photocopy
Students will be able to
- Distinguish present harms (bias, deepfakes, labor) from speculative catastrophic risk and explain why both appear in the debate.
- Build or attack a licensing plan with attention to who administers it, what it tests, and who it advantages.
- Weigh a policy against the counterfactual, not against perfection.
Vocabulary
- AI regulation — Government rules on development, deployment, and accountability.
- Existential risk (x-risk) — The argument that misaligned superhuman AI could pose civilization-scale danger.
- Alignment — Making models behave according to human values and instructions.
- Red teaming — Deliberately trying to break or trick a model before users do.
- Open weights / open models — Models whose trained parameters are published for anyone to download and run.
- Moat (competitive advantage) — Why one AI product keeps customers when models commoditize.
Agenda · 60 minutes
| Minutes | Segment | What happens |
|---|---|---|
| 0–6 | Warm-up | What do we license, and why? Does a model belong? |
| 6–18 | Background + risk map | Read the sheet. Teams sort risks into present / near / speculative, and note which a license would actually address. |
| 18–28 | Case construction | Affirmative plan: who licenses, what is tested, threshold, penalties, open-weights treatment. Negative: attacks + optional counter-plan (liability, disclosure, compute reporting). |
| 28–52 | Policy debate | Aff constructive 3 · Neg cross-ex 2 · Neg constructive 3 · Aff cross-ex 2 · Neg rebuttal 2 · Aff rebuttal 2. Judges weigh plan vs. counterfactual. |
| 52–60 | Written position | Exit ticket: policy, the risk it addresses, the risk it doesn't. |
Affirmative — license before release
We license drugs, planes, and bridges because the downside is public and irreversible; frontier models qualify.
- The developers themselves say they can't fully predict what their models will do. That's the definition of a case for testing.
- Red-teaming by an independent body finds what internal teams are paid not to find.
- Present harms — deepfakes, bias at scale — are already here; a license regime creates a body that can act.
- 'Someone else will build it' has never excused a factory from safety rules.
Negative — the plan entrenches the powerful and doesn't test for what matters
No test certifies alignment; a license regime mostly certifies who can afford lawyers — and pushes the rest offshore or underground.
- The companies lobbying for licensing are the ones who'd pass. That's a moat, not safety.
- Open-weights models can't be recalled; the rule catches responsible labs and misses the rest.
- Catastrophic risk is speculative; the regime spends its energy there while present harms go unaddressed.
- Counter-plan: strict liability for harms, mandatory disclosure, and compute reporting — accountability without gatekeeping.
Discussion questions
- 1.openingWhat makes something worth licensing? Apply your test to a model.
- 2.coreCan a pre-release test certify safety if alignment is unsolved? What would it certify instead?
- 3.coreIf the biggest firms support the rule, is that evidence for it or against it?
- 4.coreHow does your plan treat a model whose weights are already public?
- 5.closingCompare your plan to no plan — not to a perfect one. Is the world better?
How policy debate runs
The proposition is a concrete policy ("Our school should…", "The law should…"). Affirmative proposes a plan and defends it; Negative attacks the plan and may offer a counter-plan. Constructive speeches (3 min), cross-examination (2 min), rebuttals (2 min). Judges score on plan, evidence, and clash.
In grades 9–12
Stakes are real: college, jobs, creative work, civil rights. Students can steelman, cross-examine, and write policy.
Preview the full pack
Cover, teacher guide, and every student handout. Print teacher pages once and student pages one per learner.
On a phone, open the PDF in a new tab — in-page previews are unreliable in mobile Safari.
Standards
- CCSS ELA SL.9-10.1 / SL.11-12.1
- CCSS ELA W.9-10.1 / W.11-12.1 (argument with counterclaims)
- C3 Framework D2.Civ.13 / D4.7 (policy analysis, taking informed action)
- CSTA 3A-IC-24 / 3B-IC-25 (ethics and impacts of computing)
More for grades 9–12
Taught it? Tell us how it went: send classroom notes · AI concepts field guide · Try the Activities Pass free · Commission a full unit