I think it's unsettled law, although I'd be curious to know if anyone has a court case on point.

From an old JoA article:

https://www.journalofaccountancy.com/issues/2012/mar/20114558.html

"The IRS has not given guidance on how distributions from Sec. 529 plans affect the support tests. These distributions can be substantial. If the distribution is counted as support provided by the beneficiary (child), it could prevent the child from qualifying as a dependent. Sec. 529 plans allow the owner (usually a parent or grandparent) to change the beneficiary. This provides some support for the argument that Sec. 529 plan distributions should count as support from the account owner and not count as support provided by the child, but tax practitioners are still waiting for a definitive answer from the IRS."

I'd venture a guess to say the answer is muddied in state law (since the majority of 529 plans are state run plans).  I'm also thinking the answer may lie in how the money flows.  An account owner can take a distribution from the plan, then use those funds to pay the tuition of their child.  Or an account owner can direct the plan to pay the school directly.  Oddly enough the same end result (school gets paid) results in opposing positions in 1099-Q reporting.  For the former, the 1099-Q is issued in the parent's (owner's) name.  For the latter, the 1099-Q is issued to the student.

As a thought experiment this may be something of a Schrodinger's Cat problem.  The funds are both owner's and beneficiary's until the act of making a distribution determines their fate.

Rick