Motion prep
Policy · tech · Hard

Resolved: The United States federal government should substantially increase the regulation of large-language-model artificial intelligence systems.

A broad policy stem. The resolution is the topic area, not the plan, so the affirmative is whatever plan text you can defend as topical.

Runs inPolicy / Congress adaptable
DifficultyHard
SidesAffirmative / Negative
Reading the motion
What it asks

Nothing on its own. This is a topic stem: the aff writes a plan inside it. "Substantially increase" and "regulation" are the words the negative will run topicality on.

Who proves what

Aff needs a plan text, a solvency advocate, and inherency. Neg gets topicality, disads, counterplans, and kritiks. Neither side is arguing the resolution in the abstract.

Ground worth taking

Standard aff areas: pre-deployment evaluation mandates, compute-threshold licensing, training-data disclosure, liability for model outputs, and biosecurity screening on model capability. Each has a distinct disad profile, so pick one and build the block around its specific link turns.

Both cases
Affirmative

The capability curve is outrunning the evaluation regime, and only a federal mandate creates a pre-deployment check.

01 Inherency: voluntary commitments are not binding
Claim
Current safety practice rests on lab self-governance.
Warrant
The White House voluntary commitments and lab safety frameworks carry no enforcement, no defined threshold, and no penalty. A lab may revise its own policy at will, and several have.
Impact
The status quo has no mechanism that survives commercial pressure, which is the condition the plan is designed for.
02 Bioweapon uplift advantage
Claim
Frontier models measurably assist non-expert users with pathogen-related tasks.
Warrant
Red-team evaluations have found uplift on protocol synthesis relative to internet-only baselines. Screening obligations on both the model and the DNA-synthesis supply chain close the path.
Impact
Low probability, extinction-adjacent magnitude, and the plan is one of few interventions that acts before rather than after. This is the standard high-magnitude aff on this topic.
03 Modeling advantage
Claim
US federal rules become the global default.
Warrant
The Brussels effect is the precedent: firms build to the strictest large market and export that standard. The US hosts the frontier labs, so a US rule binds where the EU rule does not reach.
Impact
The plan's effect extends beyond US jurisdiction, which is how aff answers the "other countries will not comply" argument on the flow.
Negative

The plan trades off with the innovation and the state capacity it needs to work, and there is a counterplan that captures the offense without the link.

01 Innovation / China DA
Claim
Compliance burden slows US development relative to Chinese labs.
Warrant
Fixed regulatory cost falls hardest on smaller labs and open-weight developers. Uniqueness is contested and the aff will have link turns, so the block needs an internal-link story about which specific capability slips.
Impact
Military and economic leverage transfer to a competitor. Standard impact scenarios run through Taiwan or through standard-setting bodies.
02 States counterplan
Claim
Fifty states should adopt the regulation through an interstate compact.
Warrant
California and Colorado have already legislated in this space and the industry concentration in California gives a single state substantial reach.
Impact
Captures aff solvency, avoids the federalism and federal-capacity net benefits, and puts the aff on the wrong side of a "why federal" burden they often have not pre-empted.
03 Cap K / techno-solutionism
Claim
Regulating the outputs of AI leaves intact the accumulation logic that produced the harm.
Warrant
Compliance regimes are written by, and become moats for, the largest firms. The aff's reform stabilises the industry it is criticising.
Impact
The alt is refusal or a structural critique; the aff's residual harms recur under a friendlier label. Framework debate decides whether this outweighs the case.
The clash that decides it

Does federal regulation solve the capability risk, or does the states counterplan capture it while the DA turns the case?

How Affirmative wins it
Aff wins by proving a federal-only internal link: interstate compacts cannot bind labs that relocate, and export-relevant standards require federal authority.
How Negative wins it
Neg wins by consolidating the counterplan and the DA: states solve, federal action adds only the compliance cost that produces the competitiveness slip.
Where rounds go wrong
Affirmative mistakes
  • Running the resolution as the plan. You need a plan text; "regulate AI" is not one and topicality will be the whole 1NC.
  • Reading bioweapon uplift without impact calculus. Magnitude claims need probability and timeframe work or they lose to a smaller but cleaner DA.
Negative mistakes
  • Going for topicality and the case in the 2NR. Pick one and develop it.
  • Reading a generic innovation DA with no specific internal link. The aff link turn writes itself if you cannot name the capability.
Prep something next to it
This house believes AI development should be paused.
A moratorium motion where enforcement, not risk, is the actual battleground.
Parliamentary / APDA · Medium
This House would nationalise frontier AI labs.
A high-ceiling motion where the best teams argue about who ends up controlling the technology, not about whether AI is dangerous.
British Parliamentary · Hard
This house would ban billionaires.
A wealth-cap motion that turns on whether extreme fortunes are a political problem or an economic signal.
Parliamentary / APDA · Medium
This house would abolish the electoral college.
A first-principles democracy motion where the real fight is over what a presidential election is supposed to represent.
Parliamentary / APDA · Intro