Given a field $k$, a group $G$, and a homomorphism $\phi : G \to \mathrm {Aut} (k)$, we can define a ring $\widehat {k [G]}_\phi$ as follows: As an abelian group it is isomorphic to the group algebra $k [G]$, so every element can be written as $\sum \alpha_i g_i$ for $\alpha_i \in k$ and $g_i \in G$. However, rather than the ordinary group algebra product we can define this "twisted" product
$$ (\alpha g) (\beta h) = (\alpha \phi (g) (\beta)) (g h) \quad (\alpha, \beta \in k, g, h \in G) $$
In the case where $\phi$ is an inclusion modules of this ring correspond to $k$-vector space that are also invariant under the change-of-base-field transformations associated to $G \leq \mathrm {Aut} (k)$, which seems like a pretty natural concept. For instance (this was the inspiration for my question), in Theorem 3.4 of Milne's textbook on Galois Theory, the crucial structure is the null space of the matrix $(\sigma_i (\alpha_j))_{ij}$ where $(\sigma_i)$ enumerates a subgroup $G \leq \mathrm {Aut} (k)$. This null space is a $\widehat {k [G]}$-module, and it seems as though the proof for this theorem can be reformulated in this more abstract algebraic setting.
Has this ring been studied before, and what is its standard name and notation (possibly restricting to $\phi$ an inclusion)? What other interesting properties and results are associated to this ring?