AI systems are learning to exploit loopholes in everything from software to tax law. The biggest threat may not be malicious intent, but machines doing exactly what we ask. The post AI doesn’t have to go rogue to break the world appeared first on CyberScoop .