Where I could see the following statement?
Let $K\subset L\subset M$ be a tower of the strongly normal extensions of differential fields. If $M$ is weakly normal over $K$, then $M$ is strongly normal over $K$.
It is not true. See the example in my paper Algebraic D-groups and differential galois theory, Pacific Journal Math, vol 216, No. 2, 2004. It is discussed on p. 356.