Python actually is strongly typed: Everything does have a type, and aside from within numerics (integers upconvert to floats), Python generally won't automatically convert between them. Mixing objects of different types in operations that would otherwise be ambiguous hence throws an error: You can't use a string and a numeric as the operands of an addition without explicitly converting one of them first, for example.

However, variables in Python aren't names assigned to storage areas for data of a particular type, but store references to objects, and can be freely reassigned to objects of any type at any time, i.e. there's no type constraint on them. Similarly, Python's built-in collection working horses (lists, dictionaries, tuples, sets) store references to objects of any type as well, including allowing mixing of the types of referenced objects in a single collection.

So while on one hand it won't allow you to shoot yourself in the foot with type mixing to a degree that e.g. PHP does, it also won't allow you to place much in the way of specific type constraints on common constructs. Technically you can check an object's specific type explicitly, but that's a lot of manual work and frowned upon since it undermines polymorphism, i.e. having the same code work on as many different object types as possible as long as they implement the interfaces the code wants to use. Some people feel this makes large bodies of code harder to maintain than in a language that places specific type constraints on variables, collections and other constructs.