Most legal systems treat a signature as proof that someone agreed. Not proof that they understood. Not proof that the information they were given was accurate. Just proof that a pen touched paper, or a button got clicked.
That gap has always existed. It just was not visible before.
On chain attestation systems $SIGN make this more interesting to think about. The protocol creates a verifiable record that a signature happened. The record is tamper proof and timestamped. That part works.
But the record cannot capture the conditions around the signing. A document signed under pressure produces the same on chain output as one signed freely and with full understanding.
$SIGN proves execution. It does not prove intent.
That is not a criticism of the system. Every signing infrastructure has this limitation. What it does raise is a question worth sitting with. If we are building trust infrastructure at a government scale, at what point does proof of execution become enough, and when does it fall short of what trust actually requires.