You are right. The confirmed to have phrasing is awkward, and if I heard or read it I would have difficulty understanding the meaning.
The comparison with promise to is interesting. I tried various other verbs in these sentences, like said, agreed, denied, doubted. In all of the cases I thought of, the ... to have done ... phrasing sounds wrong, while ... that he had ... sounds right. But actually that's true of promise as well, if you use the same tense: promised that he had done something works, but promised to have done something does not.
So one response to your father's argument is that the to have done construction just isn't correct, even for promise.
But that isn't the whole story, because even if we switch to the tense where promised to works, most other verbs still don't sound right in that phrasing. Agreed is the only one I came up with that works this way: Chris agreed to modify the PLC tag is just fine, but Chris denied to modify the PLC tag is not.
There may be some linguistic explanation for this, but I'm not aware of it. I'd just say that for certain verbs the to do something has developed as an acceptable idiom, but for most it is not right.
(But in another context ... Chris confirmed to have modified the PLC tag would be perfectly understandable as a newspaper headline. In this case I would interpret it as Someone has confirmed that Chris modified the PLC tag, so it has a different meaning than in your example. This relies on syntactical idioms used in headline writing.)