Some more test cases: https://play.golang.org/p/Aq74lddwDH

I think this is actually working as intended.

The spec says:

A value x is assignable to a variable of type T ("x is assignable to T") in any of these cases:

• ...
• x's type V and T have identical underlying types and at least one of V or T is not a named type.

and

Each type T has an underlying type: If T is one of the predeclared boolean, numeric, or string types, or a type literal, the corresponding underlying type is T itself. Otherwise, T's underlying type is the underlying type of the type to which T refers in its type declaration.

and

Types may be named or unnamed. Named types are specified by a (possibly qualified) type name; unnamed types are specified using a type literal, which composes a new type from existing types.

Consider this code:

type T []byte
var x = []byte{1, 2, 3}
var _ T = x

T is a named type whose underlying type is []byte. X is a variable with an unnamed type (call it V) whose underlying type is also []byte. Since T and V have the same underlying type, and V is unnamed, the assignment in the third line is permitted.

Hmm.. interesting. But under this rule, shouldn't byte(4) be assignable to variable typed t5 since only one of them is named type and they both have the same underlying type byte?

@magical's explanation is correct.
To respond to @songgao's final question: no, because byte and t5 are both named types ("byte" is a name).