The main difference, of course, is that adding an ability to allow someone to intrude on privacy for a service is different than adding functionality to a product that allows someone to intrude on privacy. The service provider can act as a gatekeeper to ensure that a warrant is provided. If you make a special build that anyone can load onto a device, then, of course, there is no gatekeeper for the warrant.
The key is that I am aware of no law that requires a gatekeeper. Services ask for warrants out of respect for their customers. Often they don't respect their customers and provide access without a warrant. It's not like the software we added to the telephone switches requires some kind of special key or anything. There is no oversight and there has never been any oversight.
The thing is that (as an American in US) you are under no obligation to hand over your cell phone, unless the police have a warrant. Once they have a warrant, then I don't think you have any recourse. As we have seen, you can be found in contempt of court if you do not reveal your password anyway.
So from the perspective of the law, I think it is relatively straight forward. It's going to have to go all the way up to the supreme court. I imagine that the supreme court would choose to hear this case. Essentially you will have to argue that compelling a company to create an exploit to satisfy a warrant constitutes an unreasonable search if that exploit can also be used to perform illegal searches. I think the case would be heard, but I doubt very much that the argument will succeed. If the exploit is constructed on the condition that it is only used to satisfy the warrant in question, I'm pretty sure that it will be acceptable. We all know that such a condition need not be followed since the FBI et al don't care if the evidence they gather is admissible in court. However, I doubt it will matter legally.