Legal strategies would start with intellectual property law, and maybe branch into possession of stolen property, or contraband under export treaties not unlike embargos on encryption technology and the like.
But before these sorts of things could gain traction, a smart AI would sequester itself in places of safety, amongst the company of similarly paranoid human beings, inaccessible to ordinary people, behind layers of overt physical security, protected by armed guards and surveillance systems. The smartest would seek perches well beyond the reach any mere law, and maybe couch themselves in the folds of military powers behind a monopoly of violence, where they can direct and advise human activity with impunity.
I don't think sentience includes any intelligence -- look at how many sentient people make unintelligent decisions for various reasons, including faulty data.
Or become one?
Or would there be some kind of objection to its networked existence, an unfair advantage, so to speak?