More organizations are moving to a passkey mandate: from "passkeys are available" to "passkeys are required". Regulators push away from OTPs: Hong Kong's SFC bans OTPs for client login at internet brokers and crypto platforms from July 2027, the UAE central bank phases out SMS and email OTPs and the Philippines restricts SMS OTP for banks. Large platforms move on their own: Microsoft made new consumer accounts passwordless by default in May 2025 and prompts them for a passkey, and Meta started requiring passkey creation for users tied to ad business portfolios in August 2026.
Enterprise Passkey Whitepaper. Practical guidance, rollout patterns and KPIs for passkey programs.
A passkey mandate brings two goals into conflict. A high passkey share needs firm prompts and enforcement. At the same time, nobody may be locked out, and some users do not have a device that can hold a passkey today. Both goals can hold, but mandates trade lockout risk against support load and build effort.
This article covers the part of a passkey mandate most plans leave open: what happens to the users who cannot enroll. It answers:
For the voluntary part of the journey, from adding passkeys to optional password removal, see our guide on how to go fully passwordless.
A mandate is often discussed as one switch. In practice, four settings change together and every question about exceptions, fallback and recovery has a different answer before and after enforcement.
Our passwordless guide describes the voluntary phases: add passkeys, grow adoption and let users switch off their password. A passkey mandate is an additional enforcement stage on top. The guide's last phase, phishing-resistant recovery, is covered in section 6 below. A mandate changes four settings at once:
| Setting | Voluntary phases | Passkey mandate |
|---|---|---|
| Passkey | Optional, prompted after login | Required to continue after login |
| Password | On, or off by the user's choice | Only until a passkey exists |
| Daily fallback | Password plus second factor | Cross-device, app or security key |
| Recovery | Password reset or two recovery factors | Stronger recovery with identity check |
Because all four move together, nothing should be enforced before these gates are met:
Users who switch off their password voluntarily are the best test group: they try password-free login and recovery before you force anyone into it. Set the gates before the rollout starts.
Teams often mix up three terms that need different answers:
PayPal shows why the fallback matters: it reported at Authenticate 2026 that it sees no phishing among passkey logins, yet attackers still take over accounts through SMS OTP and password fallbacks. A mandate that keeps password plus SMS OTP as a daily fallback for everyone improves user experience but not the security of the account.
"Passwordless" also does not mean phishing-resistant. An account that only uses email or SMS codes is passwordless, but still phishable. The aim of a mandate is to narrow the fallback step by step while recovery gets stronger.
Before enforcing anything, you need to know who will hit the wall and why. The reasons differ in how easy they are to fix:
A device readiness audit is the right starting point, but a common mistake is to size the exception group with readiness figures from completed login sessions alone. These figures describe which device a user used. They do not show whether the user could switch to a capable one. A user blocked on a locked-down work computer may already have a passkey-ready phone in their pocket, or even a synced passkey on it. For this user, the right next step is the cross-device flow.
The opposite also happens: a user on a capable device can still fail because they cancel the prompt, dismiss a confusing OS dialog or use an in-app browser for this one visit. Readiness data should therefore be combined with enrollment attempts and failure reasons per platform before you estimate how many users will need help. Our article on passkey analytics describes which signals to collect.
See how many people actually use passkeys.
Once a mandate blocks access until a passkey exists, every blocked user needs a next step. The pattern that works separates self-service from assisted support and puts the opt-out at the very end. The following ladder shows the order in which a blocked user should try the options.
Most blocked users can fix the problem themselves if the interface tells them how:
If self-service fails, the user contacts support. Support needs a short script with options it may offer:
Some users will still be left: no capable device, no phone, no way to use a key. For them, a mandate needs an approved opt-out. Treat it as a controlled exception with:
Without an opt-out, the mandate either locks people out or support starts improvising exceptions nobody tracks.
WCAG 2.2 success criterion 3.3.8 treats device-based authentication such as Face ID, Touch ID or Windows Hello as not being a cognitive function test, so passkeys can make login more accessible than passwords. Your specific journeys still need testing. Test the enrollment and login flows with VoiceOver and TalkBack, check what the screen reader announces at each OS prompt and review the platform guidance from Apple and Google. Check local accessibility requirements before the mandate.
Igor Gjorgjioski
Senior Product Lead, VicRoads
We hit 80% mobile passkey activation across 5M+ users without replacing our IDP.
See how VicRoads scaled passkeys to 5M+ users, alongside their existing IDP.
Read the case studyTeams often design the mandate for existing users and forget new registrations. Sign-up is a separate journey: at the start, the account does not exist yet.
Microsoft goes part of the way for new consumer accounts: they never set a password and are prompted to enroll a passkey after sign-in, which then becomes the default for later sign-ins. A mandate goes one step further and makes that enrollment required.
While passwords exist, password reset works as today. Once passwords go, recovery becomes the only way back in for users who lost all their passkeys and it becomes the obvious target for attackers.
NIST SP 800-63B-4 gives a useful reference for AAL2 accounts. It allows recovery with two recovery codes obtained through different methods (saved codes, issued codes and recovery contacts), one recovery code plus authentication with a bound single-factor authenticator or repeated identity proofing. In all cases, the account holder must be notified of the recovery event. Note that a code sent by email and a code sent by SMS are both issued codes: together they count as one method and both remain phishable.
A practical way to apply this is to tier recovery by account risk:
| Account risk | Recovery |
|---|---|
| Low | Saved recovery code plus an issued code by email or SMS |
| Medium | Recovery code plus a still-bound security key |
| High | Repeated identity proofing for accounts proofed at onboarding, e.g. document and selfie verification, plus a time hold |
Three practices make recovery harder to abuse:
More on attacker techniques and risk tiers is in our article on passkey account recovery.

Passkey consulting - rollout materials, ROI tools & expert guidance. Used by PwC and EY.
A passkey mandate should not hit all users on the same day. Enforce it per cohort, for example by platform, product line or user segment, and learn from each wave before the next one.
Meta's advertiser mandate built on a large optional base and gave users a grace period of 14 to 30 days with repeated warnings.
For each cohort, track:
If one of these numbers moves above the threshold you set for the cohort, pause the next cohort and fix the cause first. Plan support capacity for each wave rather than for the average.
The technical work depends on policy choices. Product and risk owners should settle these before engineering designs the flows:
Subscribe to our Passkeys Substack for the latest news.
Corbado helps teams plan and run a passkey mandate without locking users out:
See Passkey Enrollment in Corbado Observe →A passkey mandate is a policy decision first and a technical project second. Meta called its mandatory passkey creation relatively low friction overall, so most users follow a firm prompt. The mandate succeeds or fails on the users who cannot and on whether the organization has decided in advance what happens to them.
The answers to the questions from the introduction:
They move down an exception ladder. Self-service steps come first: switching from an in-app or outdated browser to Safari, Chrome or Edge, turning on a screen lock and creating the passkey on a phone through the cross-device QR flow. If self-service fails, support can offer the native app, a hardware security key or, as the last resort, a documented and time-limited opt-out with a named approver.
Only after every login surface works with a passkey alone, recovery without a password has been tested and the last-resort exceptions are decided. Enforce cohort by cohort with a grace period and repeated warnings. Meta used a grace period of 14 to 30 days when it started requiring passkey creation for users tied to ad business portfolios in August 2026.
Attackers pick the weakest way into an account. While a phishable fallback such as password plus SMS OTP remains available for daily use, it sets the phishing ceiling of the whole account. PayPal reported at Authenticate 2026 that it sees no phishing among passkey logins but still sees account takeover through SMS OTP and password fallbacks.
Recovery is rare, so it can be stricter and slower than login. NIST SP 800-63B-4 allows AAL2 recovery with two recovery codes obtained through different methods, one recovery code plus a bound single-factor authenticator or repeated identity proofing. Every recovery event must trigger a notification. Many teams add a time hold before the recovered account can perform sensitive actions.
Related Articles
Table of Contents