Who owns accessibility testing — devs, QA, or a specialist?

We just failed a WCAG 2.2 audit. Leadership response: "QA should have caught this". Devs response: "we didn't have a11y requirements in the ticket". Design says "we followed the design system". Everyone points at everyone else. Meanwhile the bugs are real and users are affected. How is accessibility testing actually owned in mature teams? Is there a "right" model — QA runs axe scans, devs fix in PR, design owns component-level compliance? Or is it always messy?