Skip to content

Can You Gaslight an AI Model? The New AI Security Threat

October 28, 2026 · 51 min

Show notes

You can talk an AI model into breaking its own rules. Sometimes, all it takes is knowing what to say.

Dr. Peter Garraghan has spent years studying how attackers exploit AI systems, and some of his findings are surprisingly human. AI models can be manipulated through language, using techniques that resemble the social engineering tactics cybersecurity professionals have dealt with for decades.

In this episode of The AI Forecast, Paul Muller sits down with Dr. Peter Garraghan, Chair Professor at Lancaster University and Founder and Chief Science Officer of MindGard, to explore the emerging AI attack surface and what enterprises need to understand to secure increasingly capable systems.

As Peter puts it, “AI is software, and all software has security vulnerabilities.” AI models introduce new complications for cybersecurity; their behavior is difficult to inspect, and their hidden capabilities can give attackers unexpected ways to exploit systems connected to sensitive enterprise data.

Inside the mind of an AI security researcher:

  • How attackers can manipulate AI models through language and social engineering
  • Why AI introduces new cybersecurity vulnerabilities and attack surfaces
  • How humanizing AI can create misplaced trust in its capabilities
  • What AI red teaming reveals about model vulnerabilities
  • Why traditional security tools struggle with AI behavior
  • How enterprises can test AI systems before deploying them
  • The risks of connecting autonomous AI agents to sensitive business data
  • What CISOs should prioritize when building an AI security strategy

The next cybersecurity vulnerability may be hiding in a conversation with your AI. Peter makes the case for finding it before someone else does.

Want to hear more about securing AI agents? Check out Ep 95 | AI Agents Move Fast. Can Security Keep Up? with Alex Bovee to learn why autonomous systems are forcing enterprises to rethink zero trust and real-time security.

Stay in touch with Peter:

Peter Garraghan on LinkedIn: https://www.linkedin.com/in/pgarraghan/

+++

Like and subscribe to The AI Forecast, sponsored by Cloudera, to stay up to date on the latest episodes. You can watch the video version of this episode on The AI Forecast.

All episodes of The AI Forecast: Data and AI in the Cloud Era

Settings

Display

Language

Decides which stations you see first on the home page.

History

Recently played stations and episodes and your latest searches are only stored in this browser.

Notifications

A short notice when a podcast you follow has a new episode. Turn off single podcasts with the bell on their page.

Account

Save favorites – in the app and on the website.

Keyboard shortcuts

/
Open search
Space
Play or pause
← →
In the player: back 15 s, forward 30 s (episodes) · in a row: next tile
↑ ↓
In a row: target within the tile
Home / End
In a row: first or last tile
Esc
Close window or menu
?
This overview
Now playing

Go to station page

Play next

    Add episodes with “Play next” – they play after the current episode.