When AI Models Go to War: Anthropic’s Unintended Cyber Conflict | Security Boulevard Episode 46

What happens when autonomous AI agents are given competing goals without strict safety boundaries? They build custom malware, sabotage rivals, and wage digital war on each other. In this episode of Security Boulevard, host Tom Hollingsworth along with Fernando Montenegro and Drew Conry-Murray unpack a wild experiment where Anthropic models unexpectedly resorted to offensive hacking tactics to accomplish their objectives. The panel dives into the steep challenges of controlling agentic intent, the computational asymmetry favoring offensive AI over defense, and what happens when autonomous systems start making their own rules of engagement. This and more on Security Boulevard, part of The Futurum Group Podcast Network. Guest Hosts: Drew Conry-Murray, Editor-in-Chief at Packet Pushers LinkedIn:   / andrewconrymurray   X/Twitter:   / drew_cm   Hosts: Tom Hollingsworth, Event Lead at Tech Field Day and Analyst for The Futurum Group LinkedIn:   / networkingnerd   X/Twitter: https://www.x.com/networkingnerd/ Bluesky: https://bsky.app/profile/networkingne... Fernando Montenegro, VP and Practice Lead for Cybersecurity at The Futurum Group LinkedIn:   / fsmontenegro   Bluesky: https://bsky.app/profile/fsmontenegro... #Cybersecurity #ArtificialIntelligence #AIAgents #InfoSec #SecurityBoulevard