Full nodes derive their security from enforcing all of the rules of Bitcoin self-sufficiently on any data they receive, and by validating the complete blockchain from the genesis block to the chain tip. This does not require keeping the data after validation.
Pruning nodes have the same security that archival nodes have, the only difference is that they cannot serve the blocks that they have pruned. As they keep all blockheaders, I assume that they could even recover from a chain reorganization beyond the pruned blocks although such a thing should never happen (the minimum pruning size is 540MB). Since following Bitcoin's rules is sufficient to converge to the chain-tip, pruning nodes have full consensus capability.
If a pruning node requires a
-reindex for any reason, it will need to repeat the full Initial Block Download (IBD).