Random robots simply gone crazy, we might win. However I'm going to go ahead and think crazy singularity level AI and then we are screwed. All it has to do is launch a rocket out to like Pluto and have the rocket build Pluto into a giant computer/factory mix and we couldn't do a thing. Also you know it would be a hive mind because the sentient one would just hack all the others one by one when it became sentient. There would be no hope, they probably wouldn't go for the mass invasion crazy thing but would just take over major military and power bases. Don't think stupid robots running around with pistols, think never sleeping, never relaxing, heat seeing, motion sensing, sniper robots. Tanks can overrun snipers right? They would simply artillery shell any vehicle they see also.
Sentient AI would definitely go into crazy fail safes since it would be the only one of its race, and of course wouldn't want to die. It send minor computers into the earth with geothermal energy harvesters, launch solar powered robots into space that would never be caught.
Paradoxes and viruses would not work. If it is sentient it would simply ignore whatever nonsense you say, or if it listened simply realize its a trick.
Sentient AI processing a Paradox
Information Harmful?
Possibly Helpful?
If check send to processor 679 to be analyzed.
Processor 679 overloaded by paradox.
Processor 679 abort current command, send information leading to overload.
Overload caused by paradox, paradox is now categorized as harmful.
Viruses would be a joke, you couldn't send it a virus because it would immediately recognize commands from outside sources or, as before if infected wipe out computer effected. A hive mind computer would create virtual computers within itself so it could multitask and do damage control.
THERE IS NO HOPE THE SINGULARITY IS HERE!!!