New paper argues excessive AI provider liability weakens defense

Economist Joshua Gans of the University of Toronto has posted a new paper asking how much liability AI providers should bear when their services enable both attack and defense.

The core result is counterintuitive: liability can improve welfare while increasing harm. A higher common price reduces effort on both sides without changing attack success, but compensation weakens defense and raises attacker profits.

The paper states that under monopoly with sufficiently many productive users, zero liability remains uniquely optimal for a range of parameters. This is a theoretical SSRN preprint, not empirically tested, per Tyler Cowen's blog summary.