Amazon: AI Models Should Be Released Only When ‘Ready and Safe’

Amazon's intervention matters less for what it proposes than for who is saying it: a hyperscaler with its own model ambitions aligning with the safety-first camp rather than the accelerationist one. That shifts the political economy of AI governance, because the labs calling for caution and the platforms distributing their models are now broadly on the same side of the argument. The conspicuous gap is operational — no testing framework, no timeline — so the real question is whether Amazon's stance hardens into a concrete standard or remains a positioning statement while regulators and rivals fill in the detail.