This has been a massive bugbear of mine. Particularly when it inexplicably chooses USA date formats even when faced with a column containing values like 15-07-75. It would frequently convert half the values into US date format where possible and leave others like above unconverted.
Edit: I wonder, does '1-1' count as a 'continuous string of letters and numbers'? I still don't want '1-1' to be converted to a date.
From what I can tell, these are all opt-in via settings. So it won't stop unaware users from accidentally messing up csvs
Anyone done this? Open a .csv in excel to fix/edit a item, then save it without realizing that it autoformatted a bunch of columns. Now it doesn't work in the parent program anymore.