Yeah, to be clear I'm pro activating CTV. In terms of capability I think it's effectively equivalent to creating an ephemeral key, signing exactly one transaction with it, then destroying the key.

CTV is 64-72 bytes more efficient on chain, however, and also avoids the risks of mishandling the ephemeral keys. If they're not properly handled you could be very screwed.

Since "base" CTV can already be simulated this way, I think the risk is essentially nil. Upgraded versions that do fancier things (like what TXHASH promises) could have potential unintended consequences though, of course, and should be heavily scrutinized as they're proposed.