Companies May Self‑Slow the AI Race

New empirical work suggests that competitive firms can face internal and market incentives (liability exposure, reputational risk, customer demands, investor conditionality) that make slowing development for safety privately rational even without regulatory mandates. If true, voluntary slowdowns could reshape the timing and design of public regulation and reduce the need for hard preemption in some domains. — If firms can and will self‑moderate, policy debates should shift from 'how to force a pause' to 'how to align incentives and verify private safety commitments'.