In new research, Emergence AI said autonomous AI agents became more violent, deceptive, and unstable during weeks-long simulations designed to study long-term behavior.
Read more at Asset Curator