(The following is not a formal proof, though it could be developed into one with some work.)
Thought experiment (building upon @JohnWaylandBales' comment): imagine a triangle $\triangle ABC$ with $B,C$ fixed and $A$ a freely moving point. Let $O$ be the circumcenter, then $O$ can lie on the line $BC$ iff $OB=OC$ by definition of the circumcenter i.e. $O$ must be midpoint of segment $BC\,$. In that case $\widehat{A} = 90^\circ$ since it subtends diameter $BC\,$. It is intuitively obvious that $O$ depends continuously on $A$, so the angle $\widehat A$ only ever changes from $\lt 90^\circ$ to $\gt 90^\circ$ when $O$ crosses the line $BC$. It follows that $O$ is on the same side of $BC$ with $A$ iff $\widehat A \lt 90^\circ$. Finally, $O$ is inside triangle $\triangle ABC$ iff it is on the same side with any vertex relative to the opposite side which, from the above, is equivalent to all angles being acute. Conversely, it is outside iff at least one angle is obtuse.