The primary legal theories applied to AI liability include product liability (for defective design or manufacturing), negligence (for failure to exercise reasonable care in design, deployment, or oversight), and strict liability for inherently risky products.