Every point each debater made, checked against a source: solid or stretched.
1. "We already restrict dangerous technologies: military, uranium, pharmaceutical manufacturing."
Solid. US law licenses nuclear material, controls military exports, and registers drug makers. 10 CFR 70.3 · ITAR, 22 CFR 120 · FDA registration
2. "Open-source AI carries risks that involve the downfall of civilization."
Stretched. Today's systems "lack the capabilities to pose such risks, but they are improving", and expert opinion "varies greatly". International AI Safety Report 2026
3. "Hard to regulate is no reason not to. We regulate aviation."
Solid. Aviation runs on binding international standards under the Chicago Convention. ICAO, Article 37
4. "Perfect enforcement isn't possible, but reducing leakage still helps."
Partly solid. Weights can be pulled from platforms like Hugging Face, but once downloaded "they can share them through other means". NTIA report, 2024
1. "It's very hard to restrict open source on a global scale."
Solid. Developers "cannot rescind access to the weights", and once released "it is difficult to 'un-release' them". NTIA report, 2024
2. "Foreign actors can still use anything the US restricts."
Partly solid. True for models: China's DeepSeek released R1 openly. "Anything" overreaches, since the US does control military exports. DeepSeek-R1 · ITAR
3. "Restricting AI models limits competition."
Solid. "The potential entrenchment of incumbent firms risks significant harm to competition." NTIA report, 2024
4. "Restricting only raises costs for US consumers."
Stretched. No source found showing restrictions would raise consumer prices. Unsourced, so implied incorrect.
5. "Restrict the use, not the model. Scamming should be illegal either way."
Solid on the law. Impersonating a business is already illegal whatever tool is used. FTC rule, 16 CFR 461.3
Both debaters landed real points. Adam is right that dangerous technology is already restricted and that hard to enforce doesn't mean unregulated, but "downfall of civilization" runs ahead of the evidence. Preston is right that released weights can't be recalled and that restriction can hurt competition, but his cost claim has no evidence and "anything" overreaches.