SN 1093: Tokens in the Stream - Why LLMs are inherently insecure and prompt injection will persist

About this episode

The episode discusses the inherent insecurities of language models and the ongoing threat of prompt injection in AI systems.

Turns out, every chatbot conversation runs on a messy hack at the heart of language models, making prompt injection an unsolved—and possibly unsolvable—security threat. Steve and Leo unravel the research that explains why "roles" in AI aren't what you think they are. Understanding the controversy surrounding "AI Model Distillation" Anthropic moves to make their most powerful Mythos 5 model more widely available. Bitwarden's "Secrets Manager" offering prevents agentic and prompt injection abuse. The astounding and disturbing truth about the way conversation AI actually works Show Notes - https://www.grc.com/sn/SN-1093-Notes.pdf Hosts: Steve Gibson and Leo Laporte Download or subscribe to Security Now at https://twit.tv/shows/security-now . You can submit a question to Security Now at the GRC Feedback Page . For 16kbps versions, transcripts, and notes (including fixes), visit Steve's site: grc.com , also the home of the best disk maintenance and recovery utility ever written Spinrite 6 . Join Club TWiT for Ad-Free Podcasts! Support what you love and get ad-free audio and video feeds, a members-only Discord, and exclusive content. Join today: https://twit.tv/clubtwit Sponsors…

People in this episode

Hosts: Steve Gibson, Leo Laporte

Topics covered

• AI security
• prompt injection
• language models
• AI Model Distillation
• conversation AI
• Bitwarden

Keywords

• tokens
• security threat
• chatbot
• AI
• prompt injection
• language models
• Bitwarden
• Anthropic
• Mythos 5

Sponsors

ThreatLocker, Box, Hoxhunt, GuardSquare, Material Security

Mentioned in this episode

Organizations: Anthropic, Bitwarden, TWiT, GRC

Products: Mythos 5, Secrets Manager, Spinrite 6

More episodes of Security Now (Audio)

• SN 1095: AI-Driven Expertise Loss - Gemini, Hugging Face, and the AI Arms Race · September 9, 2026 · 3h 6m
• SN 1094: AI Patching Shortcomings - Should You Trust AI-Generated Code? · September 2, 2026 · 2h 52m
• SN 1092: Restraint Abliteration - Rotating Keys, Broken Guardrails · August 19, 2026 · 2h 49m
• SN 1091: The Post BlackHat State of AI - When AI Writes Malware · August 12, 2026 · 2h 51m
• SN 1090: Black Hat - The Hidden Flaws in AI Security Nobody Saw Coming · August 6, 2026 · 2h 5m
• SN 1089: Models Go Rogue & ExploitGym - Regulators, Start Your Engines · July 29, 2026 · 3h 8m