‘Very good’ at manipulating humans: Why smarter, deceptive AI systems pose a huge risk
Can AI systems deceive humans? The authors argue a range of systems have already learned to do this – and the risks range from fraud and election tampering to us losing control over AI.
