A pragmatic bet, not an oversight
Trust on first use accepts a credential's claimed identity the first time a system encounters it — there's no independent, trusted source to verify the claim against at that initial moment — and then holds the credential to that same identity for every use afterward. If the identity presented the first time was already an impostor's, the system has trusted the wrong party from that point forward, without ever having had a genuine opportunity to catch the mistake at all.
This is a deliberate, pragmatic tradeoff rather than a design flaw someone overlooked: it accepts a real, if generally small, risk at the single moment of first contact, in exchange for not requiring an independent verification infrastructure for every single new device or credential a system will ever encounter over its entire lifetime.
When the tradeoff is worth accepting, and when it isn't
The pattern is the right choice specifically when the first-contact risk is genuinely low, and building out full independent verification infrastructure would cost far more than the risk being accepted actually justifies — which describes a great deal of ordinary device pairing and low-stakes first connections. It's the wrong choice for anything where the consequence of trusting an impostor at that first moment would be severe enough that the cost of real, independent verification is clearly worth paying instead.