5 billion PASCIs are in use right now, and not one can be phished. So why is credential phishing still working? Because nobody's attacking the PASCI. Picture a login page that's pixel-perfect. Same fonts, same padlock, wrong domain. You sign in, and nothing happens. Your browser won't even offer the PASCI. That credential was stamped with one domain at creation, and every sign in, the browser signs over the origin you're actually on. Wrong origin, no signature. So attackers stop trying. Proof points work around. The proxy tells Microsoft Entra, you're on Safari on Windows. A combination Entra doesn't support PASCYs for. Entra turns Fido off and offers a fallback. You tap a six-digit code. The proxy takes your session cookie. The PASCY never failed. It was never invoked. Two structural holes. A relying party ID can be your root domain. So a passkey scoped there works on every subdomain. One forgotten subdomain is an auth bypass. And the private key never leaving your enclave? For synced passkeys? False. NIST bars them from AAL3 because the key is exportable. And last month, Unit 42 pulled synced keys out of Chrome's memory. The cryptography is holding up fine. It's the doors either side of it that aren't. Your passkey is only as strong as the weakest way back into your account.
Machine-transcribed from the published render, so acronyms may be cased as ordinary words.