Try returning some arbitrary value which you know is incorrect (like a list with just 1 random number), and see if you get a message like expected: your_wrong_output, but got: something_else. I think the actual/expected values are flipped, because this case in the fixed tests is correct.

Most probably, you're mutating the input and that affects the expected results.

Specify the language, please.

Looking at the kata completions I don't think there should be any issues with any of the kata versions.