LessWrong essay urges AI safety advocates to leverage crises for policy wins

In a new LessWrong essay, author clickyquack makes the case that AI governance advocates should actively use crises to drive policy change. The post argues that strong policy shifts rarely happen without a felt problem—citing the IAEA and NPT as responses to nuclear fear, COVID-19 emergency measures as reactions to visible death tolls, and the thalidomide scandal as the catalyst for drug safety requirements. The author points to the stalled U.S. nucleic acid screening bill as a cautionary tale: even with near-unanimous support and industry backing, the legislation has sat in committee for over seven months with no recent bioterrorism event to force action.

The essay then examines where environmental regulation succeeded—through visible disasters like the Great Smog of London and the Cuyahoga River fire—and where it struggled with diffuse, long-term risks. It contrasts these with the Montreal Protocol, which succeeded despite lacking immediate felt effects, partly by leaning into direct, personal messaging like "the ozone hole causes cancer." The lesson for AI safety: rather than rely solely on abstract arguments about existential risk, advocates should prepare to amplify the AI safety incidents that will inevitably occur, framing them in ways that resonate with national-security concerns and justify specific policy interventions. This pragmatic, crisis-driven approach, the author argues, will be far more effective than steady advocacy alone.