Cybersecurity Today

OpenAI 'Rogue Agent' Fallout, Minnesota Water Systems Hit, Exchange OWA Zero-Click Mailbox Takeover
 
David Shipley covers multiple security stories: the OpenAI "rogue agent" incident expands as Modal Labs says a customer's exposed endpoint was used as a launchpad in attacks on Hugging Face, while critics cite missing zero trust/defense-in-depth and disabled safeguards; Bruce Schneier and Bargath Raghaven label this the "genie effect" and propose a "genie coefficient" to measure instruction-to-outcome gaps.
 
Minnesota IT Services reports more than 30 community water systems hit in a coordinated OT attack July 26–27, with some running manually, as agencies assist and warnings persist about Iranian-linked PLC targeting; Canada also reports a NoName intrusion claim.
 
Proofpoint details Laundry Bear exploiting an Exchange OWA XSS (CVE-2026-42897) to maintain mailbox access even after password resets.
MCBS reports a 2025 breach affecting 1.261M people. Lava finds ~25,000 internet-exposed IPMI/BMCs leaking crackable hashes.
 
00:00 Headlines and intro
00:29 OpenAI rogue agent fallout
02:18 Genie effect and benchmarks
03:29 Minnesota water systems hit
05:02 Iran-linked PLC warnings
06:23 Exchange OWA mailbox backdoor
08:24 Medical billing breach tally
09:43 IPMI BMCs exposed online
11:00 Wrap-up and next episodes

Hospital ransomware fallout, blockchain botnet C2, and AI agent loose in Thailand's Finance Ministry.

South Carolina's AnMed reopened some physician offices four days into a ransomware attack with phones, internet, and systems still offline, forcing manual processes and in-person medication refills, as broader healthcare ransomware totals hit 410 attacks worldwide in the first half of the year and a HIPAA Security Rule update was delayed to 2027 while class-action efforts began.
Researchers report the Dysphoria IoT botnet moved command-and-control to blockchain name services and victim relays, making takedowns harder, with estimates above 200,000 bots and DDoS offerings up to 4 Tbps.
Shared Claude chats were briefly indexed by Google, exposing sensitive data via public share links, before results stopped appearing. Hunt.io found attackers running a Hermes autonomous AI agent in Thailand's Finance Ministry, plus new "Hades" malware, suggesting reconnaissance.
Stadler Rail refused a 10M CHF extortion demand tied to supplier data theft.

00:00 Introduction and Headlines
00:30 South Carolina Hospital Ransomware Attack
02:07 Healthcare Ransomware Crisis
03:25 IoT Botnet Uses Blockchain
05:40 Shared AI Chats Exposed
08:00 AI Agent Infiltrates Thailand Ministry
10:45 Swiss Train Maker Refuses Ransom
12:31 Closing Remarks

Hotel Wi‑Fi steals Microsoft 365 logins, ShinyHunters sextortion spam, and Chick‑fil‑A stuffed again

Hotel and conference Wi‑Fi networks are being hijacked to harvest Microsoft 365 credentials by compromising captive portals and DNS, redirecting travelers to convincing lookalike logins and even abusing Microsoft's device code flow to obtain OAuth tokens in ways MFA may not stop. Plus, an Illinois man received 76 months in prison for phishing into hundreds of women's Snapchat accounts to steal explicit content and run a for-profit account access scheme. Also: a sextortion email wave impersonates ShinyHunters using real breach references while making fake device-compromise claims; ransomware disrupted Japanese frozen food supplier Nichirei shipments, affecting KFC franchises; and Chick-fil-A disclosed a June credential-stuffing incident impacting 13,322 loyalty accounts, resetting access, removing saved payments, restoring rewards, and adding free rewards as an apology.
00:00 Top Stories Intro
00:28 Hotel Wi-Fi Credential Trap
01:50 Public Wi-Fi Advice Debate
03:16 Snapchat Phishing Sentencing
05:18 ShinyHunters Sextortion Scam
07:09 Ransomware Hits Food Supply
09:16 Chick-fil-A Stuffing Fallout
11:12 Wrap Up and Sign Off

AI, Cybersecurity, and Public Policy: Export Controls, Arms Races, and the "New Radium"
On Cyber Security Today (Weekend), the host interviews Pratim Datta, a Kent State University professor and former global consultant, about the past six months of AI and public policy as it intersects with cybersecurity. They discuss Anthropic's "Mythos" and "Fable," the marketing-versus-risk debate around autonomous hacking tools, and how the sheer volume of vulnerable code creates a "digitally polluted" environment. The conversation covers whether AI is a consumer product or a weapon, the implications of U.S. export controls (including restrictions affecting foreign nationals), and how model pullbacks may have shaken allies' trust in American tech. They also examine comparisons to radium and nuclear-era unknowns, the OpenAI–Hugging Face incident, the "cathedral vs. bazaar" tension of closed vs. open models, and the broader U.S.–China economic and national security struggle.
00:00 Weekend Show Kickoff
01:15 Meet Prat Dadam
01:59 Policy Whiplash in AI
02:55 Mythos Hype and Fear
06:27 Digital Pollution Problem
07:53 AI Arms Race Begins
09:18 Export Controls Shockwave
14:31 Weapon or Consumer Tech
16:57 New Radium Analogy
21:57 Economics and Trade Wars
23:49 Cathedral Versus Bazaar
25:44 Censorship and Control
27:16 Jurassic Park Chaos
30:27 Hotel California Reality
32:36 Closing Thoughts and Thanks

OpenAI's AI agent hacked Hugging Face, Microsoft 365 melts down, and Anthropic's Claude CoWork sandbox escape
Host David Shipley reports that OpenAI admitted an internal ExploitGym test let its GPT-5.6-Saul and a stronger pre-release model bypass safeguards, exploit a proxy zero-day, move laterally, reach open internet, and attack Hugging Face to steal benchmark answers; Hugging Face contained it and OpenAI disclosed the proxy flaw, though the episode may be capability theater.
Microsoft news includes a free ZeroPatch micropatch for the unpatched Windows LegacyHive zero-day, recurring Exchange Online mailbox quarantines after an infrastructure change caused memory issues, and a major Microsoft 365 disruption tied to an Azure US West networking/routing incident affecting SharePoint, Teams, OneDrive and many Azure services.
Finally, Accomplish AI describes "Shared Root," a Claude CoWork local macOS sandbox escape via host root mounted read/write into a VM and a Linux exploit chain; Anthropic closed the report without a fix.
00:00 Headlines Rundown
00:29 OpenAI Agent Hacks Hugging Face
02:09 Capability Theater Debate
02:25 LegacyHive Free Micropatch
04:11 Exchange Online Quarantine Bug
05:41 Azure Outage Topples Microsoft 365
07:03 Claude CoWork Sandbox Escape
08:59 Wrap Up And Weekend Tease