Security Weekly Podcast Network (Video) Routers on Trial, AI Found More Bugs - PSW #947
Oct 8, 2026
Sam Bowne, a cybersecurity educator and vulnerability researcher, joins a wide-ranging discussion on AI’s role in finding bugs, coding, cyberattacks, and risky decision-making. Topics include autonomous attack agents, open-weight models, persistent WordPress malware, spyware abuse, NetScaler and Cisco flaws, MFA bypasses, OpenSSH side channels, and the growing strain AI-generated reports puts on bug bounty teams.
AI Snips
Chapters
Transcript
Episode notes
The WordPress Backdoor That Regrows After Cleanup
- A WordPress backdoor reportedly persisted in eight locations, allowing surviving components to regenerate after cleanup.
- The panel compares it with Bugbear, which hid across networks and reinfected systems after administrators believed remediation was complete.
AI Agents Turn Attacks Into Scalable Campaigns
- Automated AI attack campaigns can scan many targets, exploit a meaningful fraction, and steal enormous quantities of data in only days.
- Tools such as Strix, Cairn, and Hermes combine discovery, exploitation, and orchestration into scalable attacks.
Use AI To Hunt Bugs In Real Firmware
- Use AI to analyze real devices and firmware, combining established tools such as Binwalk or Unblob with emulation and source-code review.
- Testing inexpensive, real-world hardware can produce more relevant experience than relying only on intentionally vulnerable applications.