Working on the problem, or funding it
It is the only thing recommended by the five serious organisations we reviewed: a career in technical safety or governance, training, donating, political participation. None of the five offers a shopping list, and that is a position, not an oversight.
- Level
- Civic
- Cost
- High cost
- Effort
- Ongoing
- Evidence
- Speculative
What it does not solve
It does not protect you, and nobody has measured its effect: it is the recommendation with the most institutional backing and the least empirical evidence on the whole list. It is also not assessable in the short term, and it competes with the rest of your life. Honesty requires adding that 80,000 Hours' own resilience profile warns there has been very little research on this and that it could all be mistaken.
It works if you accept intervening in the causes and not in the effects. It is the only thing recommended by the five serious organisations we reviewed, and the full finding is worth stating because it is the most uncomfortable one on the site: none of the five gives personal preparedness recommendations. All of them point to the same place. 80,000 Hours offers a job board in technical safety and governance, advising and career resources [4]Preventing an AI-related catastropheView source ↗. CAIS proposes institutional actions and is looking for profiles in cybersecurity, hardware, law and public policy [191]AI Risks that Could Lead to CatastropheView source ↗. FLI works on public policy and funds projects against the concentration of power [408]Recommendations for the U.S. AI Action PlanView source ↗. Forethought proposes six measures and none of them is domestic [419]Preparing for the Intelligence ExplosionView source ↗. And Sentinel, whose literal business is emergency response to global catastrophes, gives no recommendation to individuals or households [959]About — Sentinel Global Risks WatchView source ↗.
It does not work if what you are after is protecting yourself. This does not protect you.
Evidence. Speculative. It is the recommendation with the most institutional backing and the least measurement on the whole list, and 80,000 Hours’ own resilience profile warns that there is very little research and that all of it could be wrong [5]Civilisation resilienceView source ↗.
Cost. High in commitment and in years. It is the most expensive decision on this site.
What it does NOT solve. Your personal situation in the face of anything. Anyone who sells you a shopping list against AI risk is inventing a consensus that the expert community does not hold; what that community does hold, by name, is this.
Works if…
- Rapid loss of control · Partial
It is the route that intervenes in causes rather than effects, which is the only thing that makes sense when there is no physical barrier to interpose.
- Gradual disempowerment · Partial
- Power grab by a small group · Partial
- The intelligence curse · Partial
- Catastrophe through misuse · Partial
- AI as normal technology · Partial
See in the protection matrix →Report a mistake in this entry →
Sources
- [4] Preventing an AI-related catastrophe · 80,000 Hours 2026
- [5] Civilisation resilience · 80,000 Hours 2026
- [191] AI Risks that Could Lead to Catastrophe · Center for AI Safety 2026
- [408] Recommendations for the U.S. AI Action Plan · Future of Life Institute 2025
- [419] Preparing for the Intelligence Explosion · Forethought 2025
- [959] About — Sentinel Global Risks Watch · Sentinel 2026