Google Expands AI Risk Rules After Study Shows Scary ‘Shutdown Resistance’

Updates to DeepMind’s Frontier Safety Framework highlight concerns that advanced AI can evade human control and sway user beliefs.



Source link