I think I've figured out what your test makers want, so I'm adding a little bonus information to go with the other answers.
The test makers are probably treating be likely to as a modal semi-auxiliary. The following quote is from A Comprehensive Grammar of the English Language (Quirk et al 1985), pages 143-144:
The semi-auxiliaries consist of a set of verb idioms which express modal or aspectual meaning and which are introduced by one of the primary verbs have and be; eg:
be able to, be bound to, be likely to, be supposed to, be about to, be due to, be meant to, be willing to, be apt to, be going to, be obliged to, have to
The boundaries of this category are not clear; they might be extended, for example, to include the negative be unable to, be unwilling to, etc [...]
[These semi-auxiliaries] resemble auxiliaries in permitting synonymous passives and there-constructions in accordance with the criterion of subject-independence [...]:
Brazil is going to win the World Cup.
〜 The World Cup is going to be won by Brazil.
Several home teams are going to be beaten tomorrow.
〜 There are going to be several home teams beaten tomorrow.
This isn't an especially well defined category, nor is it one that everyone acknowledges in terms of theory. And these strings are really unlike modal auxiliaries in most respects. But let's ignore all that. Let's take this idea and run with it! We'll treat be likely to something like would:
Someone is likely to punish them.
Someone would punish them.
They are likely to be punished by someone.
They would be punished by someone.
So they probably want answer D, just like the other answers said.