Security Weekly Podcast Network (Video) Are you dead?, AI Hellscape, Copilot, Blue Delta, Quishing, Confer, Aaran Leyland... - SWN #546
Jan 13, 2026
The discussion kicks off with alarming findings on LLM endpoint probes and SSRF exploits. A thought-provoking segment questions if AIs can be manipulated into revealing data. Microsoft’s Copilot removal policy raises eyebrows, paired with bold predictions about an impending AI bubble burst by 2026. The podcast highlights Blue Delta’s credential-harvesting tactics and warns of dangerous QR-code attacks, or 'quishing.' Moxie Marlinspike introduces a new privacy-focused AI assistant. Lastly, a bizarre check-in app phenomenon sparks intrigue.
AI Snips
Chapters
Books
Transcript
Episode notes
LLM Endpoints Face Real-World Probing
- Large-scale probing of LLM endpoints can be ambiguous between benign researchers and real attackers.
- Organizations must assume risk and harden LLM endpoints against SSRF and similar exploits.
Let Admins And Users Remove Copilot
- Do allow IT admins to remove unwanted preinstalled AI apps via clear policies or user controls.
- Avoid forcing Copilot on users through opaque update behavior and restrictive uninstall conditions.
2026 Could Bring An AI Reckoning
- Several industry leaders predict AI-related shocks in 2026, from a market bubble to breaches tied to autonomous agents.
- These scenarios imply new governance, agent security, and scrutiny of multi-agent workflows as attack surfaces.
