This intriguing question has been put forth by Sorin Popa as a "warm-up" for a more serious question about the free flip. I'd like to know the status of the question, and if it is still open, perhaps gather some new ideas for attacking it.

 An automorphism of a $II_{1}$ factor is called *approximately inner* if it is the pointwise weak-operator limit of inner automorphisms.

Let $M$ be a type $II_{1}$ factor, and let $M\overline{\otimes}M$ denote its tensor square. 

> Question: If the flip automorphism $\phi(x \otimes y)=y \otimes x$ of $M\overline{\otimes}M$ is path connected to the identity automorphism (in $Aut(M\overline{\otimes}M)$ with the pointwise strong operator topology), does it follow that the flip is approximately inner?