Defending Against Hidden Risks of AI Tools in the Workplace - ThreatLocker (CISO Inspired Summit Manchester)

‍

Overview

In this CISO Inspired Summit masterclass, recorded in Manchester in September 2026, Eoin McGrath, Solutions Engineer at ThreatLocker, explains how AI has changed who can attack a business and how fast. Exploits that once needed a skilled attacker or a zero-day can now be produced with the help of an LLM, while staff may be pasting company data into ChatGPT and other AI tools without thinking twice.

Eoin walks through real-world examples and live demonstrations, including agentic AI being talked into building ransomware, phishing that captures MFA codes and the risks of shadow AI, before showing how application allowlisting and Zero Trust controls block unauthorised software and keep these threats contained.

Key takeaways

  • Point an AI tool at an older or proprietary piece of software and it is likely to find vulnerabilities, lowering the bar for attackers.
  • Allowlisting can stop users running new software they have created with an AI tool unless it has been approved.
  • Ringfencing limits what LLM tools can reach; in the demo, Claude could not access a ringfenced documents folder.
  • Policies can control MCP servers, child processes and how AI tools interact with other applications and networks.
  • Zero Trust cloud access ensures only managed devices with the ThreatLocker agent can reach services such as Microsoft 365, GitHub and Salesforce.

Watch the full session on YouTube: Defending against hidden risks of AI tools in the workplace