About this episode

This episode discusses the risks associated with AI agents and practical guardrails to mitigate those risks.

AI agents are useful, but they become risky when they can take action in real systems. In this episode, Tom Eston discusses recent reporting about attackers tricking Meta’s AI support chatbot into helping hijack Instagram accounts, and why that story matters far beyond social media. Tom explains practical guardrails for AI agents: read-only access first, human approval for consequential actions, separated accounts and contexts, prompt-injection awareness, least privilege, logging, monitoring, and adversarial testing for support and account recovery workflows. Special thanks to Guardsquare for sponsoring this episode! Guardsquare is the leader in mobile application security, with multi-layered protection for your Android and iOS apps. Learn more at Guardsquare.com. ** Links mentioned on the show ** Podcast: Hackers Asked Meta AI To Let Them In. It Worked https://www.404media.co/podcast-hackers-asked-meta-ai-to-let-them-in-it-worked/ The Verge summary of the Meta/Instagram AI support chatbot exploit https://www.theverge.com/tech/941179/meta-instagram-ai-support-chatbot-exploit-hacked ** Watch this episode on YouTube ** https://youtu.be/TL3MGnI4hUU ** Become a Shared Security…

People in this episode

Host: Tom Eston

Topics covered

• AI agents
• security
• social media
• technology
• risk management

Keywords

• AI agents
• security
• Instagram
• Meta
• Guardsquare
• prompt-injection
• logging
• monitoring

Sponsors

Guardsquare

Mentioned in this episode

Organizations: Meta, Instagram, The Verge

Books & works: Hackers Asked Meta AI To Let Them In. It Worked

More episodes of Shared Security Podcast

• Surveillance Pricing: When Your Data Sets the Price · July 20, 2026 · 22 min
• Signal Phishing and Russian Intelligence Targeting Messaging Apps · July 13, 2026 · 16 min
• The Supreme Court Just Put Limits on Geofence Warrants · July 6, 2026 · 15 min
• Jay Beale on Kubernetes, DEF CON, and AI Attack Paths · June 29, 2026 · 38 min
• Can the Government Shut Down Frontier AI Overnight? · June 22, 2026 · 19 min

Explore listener stats, chart rankings, contacts and more on the Shared Security Podcast podcast page.