The US government just laid down the hammer, officially recalling frontier AI models Fable 5 and Mythos 5. In Episode 46 of the DTF Cyber Podcast, Damian, Troy, and Fern dive into the chaos. Did a sophisticated prompt jailbreak a vulnerability-chaining machine into a zero-day weapon? What does this sudden regulation mean for corporate security teams, supply chain resiliency, and the blue team vs. red team arms race? We also pull back the curtain on Anthropic's quiet 30-day data retention policy changes and discuss why the entry-level SOC analyst job might be gone in a year.
Episode Timestamps:
00:00 - Fern can’t log into Anthropic
01:16 - Clearing up miscommunications: Who is blocked?
03:50 - Is this a third-party risk or supply chain resilience issue?
04:30 - The geopolitics of AI: Handcuffing US developers
06:40 - Protecting critical infrastructure from weaponized zero-days
08:45 - The risk of wrappers and pinning a business to a single model
11:22 - CEO Dario Amodei’s ironic stance on government AI agencies
14:55 - Diversifying your enterprise portfolio of AI tools
17:05 - The fine print: 30-day prompt retention and privacy traps
19:15 - RSI: Understanding Recursive Self-Improvement suppression
21:00 - Shadow IT vs. Shadow AI in corporate environments
22:00 - Corporate MSAs: Why SaaS giants don’t negotiate terms
24:56 - AI Discovery & AI Governance: Tokenizing PII traffic
26:50 - Thick clients, APIs, and enforcing MCP gateway controls
29:34 - The CISO WhatsApp groups blow up over the weekend
32:45 - Secure by Design: Is AI a lazy shortcut for secure coding?
36:50 - Skip the fundamentals? Why laziness equals a data breach
39:40 - Behavior Drift: When AI forgets what a high vulnerability is
41:50 - Is AI coming for your job? The death of Tier 1 SOC analysts
46:30 - Going to school to learn a tool vs. learning how to learn
50:17 - Episode Final Predictions: The shift to private, local models
