LessWrong: What the Ozone Hole's 39-Year-Old Treaty Teaches Us About AI Safety
LessWrong 精选 · rss · 2026-09-22
Core argument
Marking the 39th anniversary of the Montreal Protocol—the only treaty ratified by every country—the author draws an analogy between CFC/ozone governance and AI risk: both could destroy the world despite passing every known safety test, waiting for a "warning shot" may be too late, and capability in anyone's hands endangers everyone, demanding global coordination.
Historical arc
- Theory (1973-74): Postdoc Mario Molina, assigned the problem by F. Sherwood Rowland at UC Irvine, discovered CFC chlorine would catalytically destroy stratospheric ozone, with potentially sterilizing UV consequences.
- Industry pushback: CFCs were inert, safe by every existing test; industry dismissed the theory as sci-fi—echoing today's "all evals passed" arguments.
- Consumers and states first: Public boycotts gutted spray-can sales years before regulation; Oregon and other states legislated early, and a shrinking market made later bans easier.
- "The Dark Years": After the 1978 US aerosol ban, advocates faced an 8-year stalemate—no empirical evidence of ozone loss would appear for decades, model predictions swung wildly, and Rowland even published findings unfavorable to his own case for scientific integrity.
Lessons for AI
Success relied on theory-driven precaution, public pressure, and a market already in decline. AI is far harder, but the ozone story proves unprecedented global coordination is possible.
More from AGI Musings
- AI's constant caveats and follow-up questions are making researchers think harder — geoffwolfe · 2026-09-22
- Researcher Suspects AI Desk-Rejection Tools Create More Work Than They Save — IanArawjo · 2026-09-22
- Vinod Khosla: Personal AI moats are data trust and getting tasks done — brucemacv · 2026-09-22
- Halvar Flake: CS Is Turning From an Exact Field Into a Probabilistic One — soumitrashukla9 · 2026-09-22
- Epoch AI researchers: 2026 is the first year AI capabilities meaningfully hit the world — Jsevillamol · 2026-09-22
- Slop papers leave a public trail: reviewers remember, and it can sink your research career — cocoweixu · 2026-09-22