x-risklisted

# X Risk Look beyond local success and inspect downside that can compound across systems. Important work does not excuse preventable harm. 1. Define the system and time horizon. 2. List catastrophic, systemic, and slow-burn risks (AI misalignment, regulation drag, war, energy, population, asteroids/low-frequency high-impact, fragile infrastructure). 3. Separate real risk from vague anxiety. Worry enough to make bad outcomes self-unfulfilling without inventing panic. 4. Identify regulation, energy, demographic, safety, AI, infrastructure, and dependency exposures. 5. Identify who receives the benefit and who bears each downside. 6. Define prevention, containment, monitoring, rollback, recovery, and independent-review needs. 7. Set release gates and explicit stop conditions before proceeding. 8. Decide: `proceed`, `proceed with gates`, `limit or sandbox`, `pause for evidence`, or `do not proceed`. 9. **Complete** only when affected parties, credible downside, prevention/containment, monitoring, rollback, independent review, release gates, stop conditions, and a decision are explicit. **Fail-closed stop:** Recommend pause or stop when credible harm could be irreversible, controls are untested, affected parties cannot meaningfully consent, monitoring cannot detect failure in time, rollback is unrealistic, law or policy forbids the action, or evidence is too weak to bound the risk. Do not frame safeguards as bureaucracy merely because they slow delivery. Do not let pressure for