This is happening because the checker is analyzing the string inside the annotation (three strings inside Literal). The ranges aren't correct in that case. We need to find a holistic solution to this because it's happening to all rules working on strings.

For example, here UP025 is triggered for the u'\\n' which is inside the string type annotation.

/Users/dhruv/playground/ruff/src/UP025.py:1:12: UP025 [*] Remove unicode literals from strings
  |
1 | NEWLINE_SEQUENCE: "te.Literal[u'\\n', '\\r\\n', '\\r']" = "\n"
  |            ^^^^^ UP025
  |
  = help: Remove unicode prefix

ℹ Safe fix
1   |-NEWLINE_SEQUENCE: "te.Literal[u'\\n', '\\r\\n', '\\r']" = "\n"
  1 |+NEWLINE_SEQENCE: "te.Literal[u'\\n', '\\r\\n', '\\r']" = "\n"

Found 1 error.
[*] 1 fixable with the --fix option.

We could potentially fix the range for the rules to be useful in the context of type annotation or ignore them completely. Nevertheless, we should completely avoid rules which might update the quotes because otherwise we need the context of the quote which contains the actual type annotation.

Originally posted by @dhruvmanila in #10546 (comment)