I moved to Saudi Arabia and picked up a new number. A few weeks later, setting up a TikTok account for a work project, I found one already existed on that number. Clearly active. No way to reach whoever it belonged to.

Small friction on the surface. But it pointed at a bigger question: what actually happens when a carrier reissues a number, and who owns the problem when the old account never gets flagged?

I mapped it as a product problem, not a support ticket, using STC as a public example carrier since I have no insider access to their systems.

No expiry logic tied to number reassignment. Carriers track number reuse; most downstream apps don't.

No differentiated response. The new user gets the same generic error whether the block is fraud-related or just a stale number.

No owner-verification path. There's no mechanism to confirm the old account is actually still the same person.

No risk tiering. A locked social media signup and a locked bank account get treated the same way today, when the actual risk is nowhere close to equal.

No escalation trigger. The issue never routes anywhere; it just sits.

Anything touching identity or financial risk goes to a human. No exceptions, regardless of how confident the AI's match is. The cost of a false positive here (locking out the real owner) is asymmetric with the cost of a false negative (a slightly slower resolution for a low-risk case). AI can reduce friction on the low-risk side. It should never be the final word on the high-risk side.

This isn't really about phone numbers. It's about a pattern that shows up everywhere identity gets carried across systems that don't talk to each other: a decision made once, applied automatically forever, with no mechanism to notice when the underlying facts have changed. The fix isn't more automation. It's knowing exactly where automation should stop.