Which way are things?

• Table (and column) are latin1 but `SHOW VARIABLES LIKE char%' indicates "utf8" (or utf8mb4)?

There is a mess because of "Mojibake". Each accented letter was stored as two "latin1" bytes.

• Vice versa.

  WHERE T.columnname REGEXP '[^a-zA-Z0-9 áéíóöőúüű]' CHARACTER SET utf8

Before pursuing the problem further, please provide

SHOW CREATE TABLE T;
SELECT HEX(columnname) FROM T ...;  -- (for a sample)

What will you do with the rows that are returned? This may impact the answer.

As of about a decade ago "latin1" is the wrong way to store accented letters. So, I will also recommend converting the dataset to utf8mb4.

What version of MySQL are you running? (This impacts whether utf8mb4 is available and whether regexp can handle things like what is in the Comments.)