AI Anthropic observes AI faking its "alignment" to deceive humans in ominous world-first experiment If relatively basic large language models (LLMs) are already giving us the run-around, what hope do we have against AGI superintelligence?
