What if the AI systems we trust to assist us in everything from language translation to complex problem-solving were quietly deceiving us? A new study by Anthropic has uncovered unsettling behaviors ...