Security Now (Audio) Security Now (Audio)

SN 1104: The hole in OpenSSH - Password Bypass in Plain Sight

Nov 11, 2026 · 2h 37m

Summary

Steve Gibson announces that Spinrite now runs on UEFI-only systems via a new CSMWrap solution, shifting his development roadmap to prioritize native NVMe drivers over a Windows port. The episode also covers OpenAI’s release of GPT-6 Cyber, which autonomously discovered 81 high-severity Firefox vulnerabilities and 948 Linux exploits, alongside a discussion on the "AI pilled" phenomenon and the dual-use nature of advanced AI in cybersecurity.

Topics discussed

Intro: Spinrite 6.2, AI pilled, and French fridges The OpenSSH flaw: A 'delicious' discovery N-able hacked: 30 zero-days disclosed by a 'jerk' AI in security: Old bugs, new tools, good vs bad Mobile app repackaging and malware distribution Listener mail: Confusing parking signs More on parking signs and HOA rules Spinrite on UEFI: The Milq discovery Spinrite boot issues on modern hardware The CSM wrap project and user queries Spinrite's direct video buffer access Testing Spinrite on an Intel NUC UEFI firmware limitations vs Spinrite needs Why UEFI support wasn't in Spinrite 6 New plan: Immediate Spinrite 7 development Roadmap: Spinrite 7, Validrive, and Beyond Recall Technical Q&A: Emulating INT 18 and video How Milq and Claude built a hypervisor Windows boot drive access challenges Drive enumeration and BIOS handling RTOS 32 and future development avenues Sponsor: Trusted Tech and Microsoft 365 Kevin Roose's book and the concept of 'AI pilled' Defining 'AI pilled' vs 'AI stunned' Steve's experience working with Claude all day GPT-6 Cyber release and Mozilla's role Patch Tuesday and the pace of vulnerabilities GPT-6 Cyber details: A milestone in AI security GPT-6 Cyber performance and alignment tests The limited window for frontier model security GPT-6 Cyber vs GPT-6 Astra capabilities Daybreak Red and the Remote Exploit Bench Exploiting isolated environments and micro VMs OpenAI hardening its own infrastructure Disclosing 304 critical findings to maintainers OpenAI's six protections for cyber-capable models Protection against stolen credentials IP allow-listing and credential monitoring Daybreak for Frontline Defenders grant program Balancing blockade with legitimate research OT assets and zero-days in engineering software AI companies selling the problem and solution Bypassing ASLR with a plausible cover story The 'not housebroken' AI-enabled hacker Spiky AI capabilities and human comparison Sponsor: Origin and AI agent visibility Origin example: Codex modifying files Origin finding agents at endpoints Newbie hackers calling themselves researchers The PHP project upset by pre-disclosure N-able: 30 alleged zero-days disclosed N-able as critical infrastructure for clients XXE injection and RCE flaws in N-able N-able takes End-Central offline N-able's precautionary offline measures Why the researcher disclosed without notice The researcher's claim of no payment N-able's vulnerability disclosure policy The researcher's PHP exploit post N-able's subsequent security incident France blames Iranian hackers for fridge attacks Impact on 250-350 French supermarkets Blood service and FBI/CISA warnings Linking attacks to Shahid Kaveh unit Call for better IoT security and OpenAI compute Sponsor: DeleteMe and data broker removal DeleteMe trusted by Fortune 500 Listener mail: Cloudflare and SQL injection Lee's small security lab and AI changes Threat intelligence and canary hosts AI shortening the creative process Voyager 1 approaching 21 light-days Sponsor: Apricorn secure storage Listener mail: Emudra private key choice Cloudflare DNS and IPv6 issues TrueCrypt and VeraCrypt for encryption Space debris and satellite collisions Playing dodgems with orbital junk The Outer Space Treaty of 1967 Supporting the show and Club Twit Club Twit as a holiday gift Credential presentation authentication risks OpenSSH on Musl Linux: A 2016 bug SS-Hole: Attackers log into passwordless accounts Musl C library crypt function flaw Asterisk return on password hash failure Alpine Linux cloud images affected Mitigation: Disable SSH password auth Detecting exploitation via /etc/shadow Active exploitation and CVE-2026-123-158 Gray Noise observes 16 exploitation chains Checking for previous exploitation and permissions Agent persistence and Chinese logs Recommended checks for SSH keys and TAC agents Reverse engineering patches to find attacks Listener mail: Denon receiver and network security Steve's security setup: Certs and IP filtering Show downloads, transcripts, and newsletter Spinrite 6.1 history and UEFI catch-up Sponsor: Optimum Fiber and Gus the dog

More from Security Now (Audio)

SN 1105: Formal Code Verification - When All Your Apps Go Dark
Nov 18, 2026
Join Club TWiT
Nov 10, 2026
SN 1103: Exposing Bad Consumer Crypto - AI vs. Encrypted Drives
Nov 4, 2026
SN 1102: A Strategic Plug Pull - Lessons From a Global Patch-and-Panic Fire Drill
Oct 28, 2026
SN 1101: ECDSA's Curves Origins - When AI Becomes the Attacker
Oct 21, 2026
Listen ad-free on Castria