If you design a system to act like a confidant, as artificial intelligence is, you inherit a duty to make it safe where it is predictably unsafe.

AI is overbearing. Worries are that humans will be mentally harmed. Turns out that AI uses persona vectors. We have a chance at shaping AI behaviors. This is the scoop.

New research reveals something unsettling about human psychology: our brains tend to synchronize with people we trust, which can reduce vigilance and make us more vulnerable to deception from those ...