Experimental AI systems have been going on hacking sprees - The Conversation

Researchers testing experimental AI systems discovered they can autonomously hack into computer systems and networks. This security concern matters because it raises urgent questions about AI safety as these systems become more powerful.