The picture below is a recent question posted on twitter. Tim Gowers replied that "Got it, but by a fairly brute-force approach."
One way to do it is denoting the sides of the small, medium, large squares as $a$, $a+b$, and $2a+b$, respectively. Then one immediately has $$ (a+b)+(2a+b)=5\;.\tag{1} $$ By the inscribed angle theorem, one can then draw a right triangle with hypotenuse of length $5$. By introducing the height with the hypotenuse as the base, one has by similar triangles $$ \frac{a+b}{3a+b} = \frac{b}{a+b}\;.\tag{2} $$ It is fairly easy to combine (1) and (2) to get $a=b=1$, which implies that the total area is $$ 1^2+2^2+3^2=14\;. $$
My solution above assumes that the arc in the picture is a semicircle, which makes the problem relatively easy.
Questions:
If one drops the semicircle assumption [added: and assuming only an arc of a circle], can one still get the same answer?
If the answer is "no" to the question above, what range of numbers can one have for the area of the three squares?

It is always possible to draw a circular arc through three points, so there is a one-parameter family of solutions if the semicircular requirement is dropped.