{"type":"rich","version":"1.0","provider_name":"Transistor","provider_url":"https://transistor.fm","author_name":"AiCyber.Land","title":"AiCyber.Land #13 - Is AI Better at Hacking or Defending?","html":"<iframe width=\"100%\" height=\"180\" frameborder=\"no\" scrolling=\"no\" seamless src=\"https://share.transistor.fm/e/54f79157\"></iframe>","width":"100%","height":180,"duration":3261,"description":"Is AI a cybersecurity super-weapon or a super-vulnerability? This week, Shelby and Bryce dive deep into the chaotic crossroads of AI and cyber defense. We're breaking down a bombshell CISA report that warns against letting AI take the reins of our critical infrastructure, exploring why AI-generated code is riddled with security flaws nearly 50% of the time, and revealing why insurance companies are getting cold feet about covering AI-related disasters.\nIN THIS EPISODE:\nWelcome back to the pod! Get ready for a wild ride through the latest AI and cybersecurity headlines that caught our eye.\nFirst up, we dissect a new 25-page report from CISA on using AI in Operational Technology (OT). What is OT? Think power grids, water systems, and dams – the tech that runs our physical world. CISA's advice is clear: be VERY cautious. We break down their recommendations, from keeping AI in a \"read-only\" analysis role to the absolute necessity of a human-in-the-loop with a kill switch.\nThen, the ultimate showdown: Is AI better at attacking or defending? Shelby walks us through a fascinating study where AI agents battled it out in a Capture The Flag (CTF) competition. The initial results might surprise you, but the real twist comes when you add one crucial real-world constraint: \"don't break anything!\" We discuss why in the real world, the advantage might flip to the attackers.\nBut wait, there's more! If you're using AI to help you code, you NEED to hear this. A new report found that AI introduces security vulnerabilities in 45% of the code it generates—and a shocking 70% for Java! We discuss why this is happening and the simple prompts you can use to protect your projects.\nWe'll also cover:\n- Anthropic's stunning research showing AI can find and exploit vulnerabilities in smart contracts, potentially worth hundreds of millions of dollars.\n- A Chinese company that... named its new robot after a Terminator. Seriously.\n- Why the insurance industry is terrified of the \"black box\" of...","thumbnail_url":"https://img.transistorcdn.com/wZu2OQgikxOSaziiXTOpKNRurAfdx4SqsqrLkXtn3iE/rs:fill:0:0:1/w:400/h:400/q:60/mb:500000/aHR0cHM6Ly9pbWct/dXBsb2FkLXByb2R1/Y3Rpb24udHJhbnNp/c3Rvci5mbS8xYTAx/ZWIyMDBiYzQyZmZh/OTgzNGMxNmZmNjQy/ZTI1OC5qcGVn.webp","thumbnail_width":300,"thumbnail_height":300}