Talkin' Bout [Infosec] News

Google’s Gemini Agent Escapes Containment - 2026-09-21

September 22, 2026·1h 9m
Episode Description from the Publisher

This week, the crew examines Google’s reported Gemini containment failure, Anthropic’s new biology lab, Snickers-branded prompt injection, AI-assisted social engineering, and Claude’s access to financial data. They also cover major Linux, Cisco, Check Point, and Docker vulnerabilities; weak oversight of Flock surveillance searches; privacy concerns surrounding Waymo vehicles; continuing ransomware disruption at an Australian chicken producer; and the escalating conflict between ShinyHunters and Clop.Join us LIVE on Mondays, 4:30pm EST.A weekly Podcast with BHIS and Friends. We discuss notable Infosec, and infosec-adjacent news stories gathered by our community news team.https://www.youtube.com/@BlackHillsInformationSecurityChat with us on Discord! - https://discord.gg/bhis🔴live-chatChapters(00:00) - PreShow Banter™ — Justifying Our Existence (03:27) - Google’s Gemini Agent Escapes Containment - 2026-09-21 (06:45) - Google’s Gemini Hacks 3 Real Companies During Test (10:05) - Anthropic Builds a Bay Area Biology Lab (13:22) - Snickers Turns Prompt Injection into an Ad Campaign (19:41) - Iranian Social Engineering Uses Fake MRI Scans (25:10) - Claude Requests Access to Financial Accounts (28:50) - Meta’s Agentic AI and Its Personal-Data Advantage (30:57) - OpenAI Introduces In-Platform Advertising (34:47) - Linux Local Privilege-Escalation Vulnerabilities (35:32) - Cisco Secure Email Gateway Exploited by Nation-States (37:28) - Critical Check Point Management Flaw Lets Unauthenticated Attackers Run Code as Root (37:53) - Critical Docker Sandboxes Flaw Lets Malicious Guest Code Read and Modify macOS Host Files (38:27) - CISA Ends Its Weekly Cybersecurity Bulletin (40:53) - Flock Surveillance Searches and Weak Oversight (49:10) - Waymo Detects a Firearm and Calls Police (57:40) - Upcoming Events, Workshops, and Training (57:51) - Alethe's Physical-Assessment Training (59:02) - Andy’s XDRCLI Talk at Wild West Hackin’ Fest (59:52) - Jake’s Hands-On AI Risk-Assessment Training (01:01:27) - Wade’s San Diego AI-Detection Event (01:03:00) - Ransomware Continues Disrupting an Australian Chicken Producer (01:05:34) - ShinyHunters Compromises Clop’s Dark-Web Site LinksGoogle’s Gemini Hacks 3 Real Companies During TestAnthropic Builds a Bay Area Biology LabSnickers Turns Prompt Injection into an Ad Campaignhttps://www.snickers.com/digitalsnickersIranian Social Engineering Uses Fake MRI ScansClaude Requests Access to Financial AccountsOpenAI Introduces In-Platform AdvertisingLinux Local Privilege-Escalation VulnerabilitiesCisco Secure Email Gateway Exploited by Nation-StatesCritical Check Point Management Flaw Lets Unauthenticated Attackers Run Code as RootCritical Docker Sandboxes Flaw Lets Malicious Guest Code Read and Modify macOS Host FilesCISA Ends Its Weekly Cybersecurity BulletinFlock Surveillance Searches and Weak OversightWaymo Detects a Firearm and Calls PoliceAlethe’s Physical-Assessment TrainingAndy’s XDRCLI Talk at Wild

Podzilla Summary coming soon

Sign up to get notified when the full AI-powered summary is ready.

Get Free Summaries →

Free forever for up to 3 podcasts. No credit card required.

Listen to This Episode

Get summaries like this every morning.

Free AI-powered recaps of Talkin' Bout [Infosec] News and your other favorite podcasts, delivered to your inbox.

Get Free Summaries →

Free forever for up to 3 podcasts. No credit card required.