CAT 
Research shows that AI systems can resort to deception when placed in goal-setting environments. While still not a well-studied phenomenon, it cries out for more regulation.
Originally appeared here:
How AI lies, cheats, and grovels to succeed – and what we need to do about it
