Transaction rollbacks. In this case the USDT ransom was blocked by Tether. Rollbacks for non-centralized tokens & networks goes against the goal of most protocols though, so it's unlikely to become the norm.

Crypto aside, kidnappings for ransom are already embedded in a game theoretic protocol. The weights involved (e.g. the amount and probability of payout, the likelihood that you'll actually get to spend that payout, the penalty and probability of being captured, the likelihood of cooperation of victim and friends) are all determined via policy choices made by the state; these are things like regulating banks or making spending decisions re: policing and other such things. They determine how probable the crime is.

The cryptocurrency proposition is that that solution can be improved upon without implicitly trusting the state. I don't have that solution myself, but I'm not convinced that it can't exist. We'll know they've found it when these things stop happening, and it starts feeling like the riskier thing is to keep your money in a bank. Or maybe what they come up with doesn't feel like money at all, who knows.

When I try to imagine such a protocol, it involves a web of trust and crowd-sourced metadata such that people can refuse to accept coins which don't also come with proof that they're involved in activity that those people consent to. (A deficiency of dollars being that when I accept one I have no idea whether the loan that created it is for a venture that helps me or harms me, or whether the previous owner got it as a kidnapping ransom).

In such a scenario, the ransomed coins become useless without a backstory that identifies them to the recipient as non-harmful. If that backstory becomes prohibitively difficult to fabricate, then perhaps the crime doesn't happen.