Lint/LiteralInInterpolation autocorrects literals inside string like nodes to remove the interpolation. However, this does not make sense for arrays inside a regexp, because the array is converted to a character class that is probably not expected.

For example:

/#{%w[a b c]}/

# compiles to... 
/["a", "b", "c"]/

# but it is much more likely the programmer wanted this instead
/[abc]/

The character class created also adds multiple quotes, spaces and commas depending on the array, which means that the rubocop correction of /#{%w[a b c]}/ becomes /[\"a, bc]/, which looks very different than where it started. As such, even if the programmer was trying to match the array literally, the resulting regexp will not do that.

I would like to suggest removing the ability for Lint/LiteralInInterpolation to deal with arrays inside a regexp, and instead add a new cop to detect literal arrays inside a regexp (and possibly correct them to proper character classes or alternation).