How do you confirm sensitive actions for SSO users without a local password?

November 16, 2026 4 answers
Asked by _d84+ferrulepiston.6

My app currently asks users to re-enter their password before changing security settings or performing destructive actions. I'm adding Google SSO, so some users won't have a local password.

Would you use another SSO login, MFA, or an email code? I'm concerned that an existing identity provider session could make a repeat login silent and provide no fresh proof.

Answers

m.382+draincompass

Start with a fresh authentication request to the identity provider, rather than adding a separate app password. A normal SSO redirect isn't enough by itself if the provider can silently reuse the existing session.

For OpenID Connect, look at max_age=0 or prompt=login, then validate the returned ID token and check auth_time against the freshness window you require. Check Google's current documentation and test the behavior before relying on either option. Provider support and the proof the user actually has to provide can vary.

Also make sure the returning identity matches the signed-in account, and tie the confirmation to the action and session that requested it. Fresh authentication doesn't automatically mean MFA, so request or enforce a stronger method separately if the action needs it.

_x92!drainruler.4:

That distinction matters. A recent authentication time tells you when authentication happened, not necessarily whether a second factor was used.

Reply to this answer
_q83~drumgauge.5

If these actions need stronger proof, look at step-up authentication with the provider's MFA policies, or add an app-side method such as WebAuthn. That gives you a different option when the provider can't deliver the assurance the action needs.

I wouldn't use the existing security question for this. And a new app password gives SSO users another credential to manage, which rather misses the point.

Reply to this answer
p.916~bitstencil

A short-lived email code is a simpler confirmation flow and avoids sending someone back through SSO every time. It can be a reasonable usability tradeoff for some actions, but don't treat it as equivalent to strong re-authentication.

If the mailbox is the same account used for SSO, the two checks may depend on the same underlying access. Consider what access an attacker would need to receive the code, especially for changes to bank details or account recovery settings.

5n~buttonladder_283:

The acceptable tradeoff may differ between a reversible change and one that moves money or locks the user out.

Reply to this answer
4m+cachepulley_759

For some actions, confirmation doesn't have to mean another login. You could put a request into a pending state and require an administrator to approve it before it takes effect. That won't fit every self-service flow, but it can work when another person is already responsible for authorizing the change.

_d84+ferrulepiston.6:

That makes sense for approval workflows. The actions I'm considering need the user to confirm them directly, though.

Reply to this answer

Add your answer

Answers are reviewed before publication.

Related Discussions