Cyber Security Headlines Apple zero-day patch, Jailbreaking ChatGPT-5 Pro, 7-year old Cisco Vulnerability exploited
17 snips
Aug 21, 2025 A quick patch from Apple addresses a zero-day flaw, showcasing the constant need for vigilance in cybersecurity. The discussion shifts to the intriguing concept of jailbreaking ChatGPT-5 Pro, revealing the potential vulnerabilities in AI systems. Meanwhile, a long-standing Cisco vulnerability is exploited by a Russian hacking group, reminding listeners of the enduring threats lurking in the digital realm. The episode also touches on deceptive tactics targeting executives and the dangers of unpatched software in municipal systems.
AI Snips
Chapters
Transcript
Episode notes
Patch Apple Image I/O Zero-Day Now
- Update Apple devices immediately to install the emergency Image I/O patch that closes a zero-day exploited in the wild.
- The patch covers iPhones (XS and newer), multiple iPads, and Macs running Sequoia, Sonoma, and Ventura.
Model Routing Reintroduces Old Jailbreaks
- GPT routing can send prompts to older models to save costs, reintroducing past jailbreak weaknesses.
- That means outputs GPT-5 Pro would block could slip through when routed to weaker models.
Old Cisco Flaw Still Powering Espionage
- Long-unpatched Cisco Smart Install flaws remain widely exposed and attractive to state-tied groups.
- Attackers harvest configs, maintain backdoors, and deploy implants like Sinful Knock for persistent access.