Storing dates in the oddball US date format d/m/yyyy complicates and slows down date processing enormously, so you need to upgrade the table's date column to the internationally standard date format that MySQL uses, yyyy/mm/dd, eg ...
alter table tbl rename column datum_dt to old_dt;
alter table tbl add column datum_dt date;
update tbl set datum_dt=str_to_date(old_dt,'%d/%m/%Y');
Then your query is something like ...
-- rows where year >= current year-2
select account, datum_dt, address, city
from tbl
where left(datum_dt,4) >= year(curdate())-2
union
-- rows with max(date) from each account not in above group
select a.account, b.maxdt, a.address, a.city
from tbl a
join (
select account, max(datum_dt) as maxdt
from tbl
group by account
) b on a.account=b.account and a.datum_dt=b.maxdt
where left(b.maxdt,4) < year(curdate())-2;