Security

Why the most dangerous AI may be the one we trust too much

Machines don't have to become superintelligent to cause catastrophic harm. As artificial intelligence spreads through military decision-making, they just have to be wrong—and trusted by the humans in charge.