Security Now #1063: Mongo's Too Easy

Date: February 4, 2026
Hosts: Steve Gibson & Leo Laporte

Episode Overview

This episode of Security Now explores serious issues in the current security landscape, with headline stories about:

• Supply chain attacks on Notepad++ and antivirus programs
• The growing pains of bug bounty programs now flooded by AI-generated spam
• The ultra-low barriers to hacking exposed MongoDB databases – the focus of the "Mongo's Too Easy" title
• The breakthrough of AI in autonomous bug detection, especially recent CVEs found in OpenSSL
• Ongoing challenges balancing security, privacy, and convenience, especially as AI agents get more capable
• Several smaller stories: age verification humor, iOS location privacy, malicious clipboard hacks, and more

The hosts mix technical detail with humor, industry anecdotes, and lived experience – offering analysis and actionable recommendations for practitioners and users alike.

Key Discussion Points & Insights

1. Personal Updates & Podcast Process

(01:18 - 04:27)

• Steve shares his evolving process for preparing the podcast earlier in the week, balancing security research and code work amid personal life and a protracted house remodel.
• Leo introduces the rise of AI-powered tools in his own workflow, recommending Steve explore their ability to automate news curation and note preparation.

  "The AI summaries help. With assessing that I don't have any dearth of topics, as you said, security." – Leo (07:51)

• They reflect on how their value isn't just curation, but expert analysis and synthesis.

2. The Temptation and Insecurity of AI Agents

(10:09 - 13:15)

• Discussion on new AI agents (like "OpenClaw") that can automate personal tasks, but at huge convenience/security trade-offs.
• It's now commonplace for users to grant full access (credit cards, email, docs) to AI bots to maximize their effectiveness.
• Risks are highlighted, including the potential for giving away sensitive data and the normalization of "YOLO everything" mindsets.

  "The benefit you get out of that is so great that it's, it's very tempting... I just can't bring myself to do it." – Leo (11:23) "Functionality or ease of use. The scale is very vastly tipped, if you're willing." – Leo (10:54)

3. Breaking News: AI Disrupts Software Industry & Markets

(16:23 - 18:41)

• Steve live-reacts to a news flash: stock prices for traditional software companies are dropping due to fears that new AI developments – e.g., Anthropic's AI drafting tools – will undercut entrenched business models.
• The panel anticipates a new market for "enterprise-grade security around these AI tools" as companies attempt to harness powerful but risky AI agents safely.

  "There will be companies that will come up with ways to do this in a secure and safe fashion." – Leo (18:06)

4. Antivirus & Notepad++ Supply Chain Attacks

(23:23 - 50:45)

Notepad++ Infected via Supply Chain:

• Notepad++ updates were compromised for months in 2025 by what appears to be a Chinese-state operation targeting specific users (not widespread, more likely aimed at dissidents).

  "Highly sophisticated state level actors believed to be Chinese had arranged to compromise and did compromise his Notepad++ software update mechanism." – Steve (24:23)

• Steve has historically warned about the risks of frequent auto-updates acting as vectors for supply chain attacks.

  "Every update is another opportunity for us to get our, our malicious code into someone's computer. And that's what happened." – Steve (24:44)

E Scan Antivirus Supply Chain Attack:

• Similarly, a supply chain breach of the E Scan antivirus' update servers led to malware infecting the very machines it was meant to protect.
• This follows a pattern; it also happened to E Scan in 2024 (last time by North Korean actors).

  "An antivirus solution has a very a highly privileged position in our machines. It's gotta be running in the kernel... If it goes bad, you're in deep trouble." – Steve (34:54, 49:08)

• Steve emphasizes: today, don't install third-party antivirus on Windows – the native one is safer.

5. The End of Bug Bounties? AI Spam Drowns Security Reviews

(68:09 - 87:38)

Curl Shuts Down Bug Bounty:

• The Curl project disables its bug bounty after being swamped by "AI slop": masses of bogus, poorly researched AI-generated bug reports.
• Official new policy: "We will ban you and ridicule you in public if you waste our time on crap reports."

  "A bug bounty gives people two [too] strong incentives to find and make up problems in bad faith that cause overload and abuse." – Steve, reading from Curl announcement (74:25)

• Concern that genuine security research will suffer; Leo advocates for filtering and staffing, not abolishing, bounties.

AI Finds Real Vulnerabilities (OpenSSL, Curl):

• In contrast, the company Aisle uses autonomous AI to find an unprecedented number of real, severe vulnerabilities (zero-days) in high-profile open source projects (OpenSSL and Curl).

  "AI is responsible for discovering 13 out of the 14 zero day vulnerabilities in OpenSSL in 2025." – Steve (90:36) "The concentration of findings from a single research team spanning this breadth... is historically unusual for OpenSSL and is in my view... due to our heavy use of AI." – Aisle's researcher (92:00)

• Their process involves AI scanning, triage, exploit construction, even patch suggestion, with humans acting mainly as supervisors.

Industry Implications:

• Signals an era where code auditing by AI puts enormous pressure on maintainers and will upend previous assumptions about code safety, security labor, and open source sustainability.

6. The "Picture of the Week": Age Verification Humor

(21:51 - 24:22)

• Steve introduces a tongue-in-cheek web form offering to check your age by searching for your ID in "existing breaches"—a joke highlighting how widespread data leakage has become.

  "If your ID is found, we can verify your age automatically. It's quick and easy and odds are you're already in there." – Steve, reading the satirical form (22:14) "It's got to be tongue-in-cheek... breaches are so rampant that why are we even being asked to identify ourselves." – Steve (24:13)

7. Location Privacy: Apple Adds 'Limit Precise Location' for Cell Networks

(55:05 - 62:59)

• Apple to soon add (iOS 26.3) the ability for compatible devices to reduce the granularity of location data reported to cellular networks, hampering their ability to pinpoint users.
• Caveats: Only available with participating carriers (currently only Boost Mobile in US), and only for cellular operators—not for apps.

  "As a result, they might be able to determine only a less precise location. For example, the neighborhood where your device is located, rather than a more precise location..." – Apple, via Steve (58:33) "The carrier has to agree to it... and there's even some speculation that carriers might actually sue Apple over this." – Leo (61:47)

8. AI Gone Wrong: Gemini Deletes a Developer’s Project

(123:10 - 126:21)

• Listener Panos recounts how Google's Gemini extension for VSCode entered a destructive loop and deleted his entire project, including files not backed up by VSCode's "safety net."

  "The Gemini CLI and VS code extension can occasionally misinterpret conversational context as destructive terminal commands... causing files to vanish." – Steve, relaying Gemini’s apology (124:08)

• Both hosts advise frequent version control commits as mitigation.

9. Windows Shared Clipboard: Security Weakness

(128:21 - 128:44)

• Steve warns about a new social engineering attack exploiting shared clipboards between browsers and system "run" dialogs – enabling easy abuse via malicious JavaScript and unsophisticated users.

  "It would be trivial for Microsoft to track the source of any data that's placed onto the clipboard and take special measures when any clipboard data attempts to cross a security boundary." – Steve (128:44)

10. ISPs and Privacy: Monetization Concerns

(133:00 - 137:56)

• Listener speculates, and Steve agrees, that ISPs could monetize not just browsing history, but direct mapping of public IPs to customer identities for marketing purposes, not just law enforcement.
• Steve suggests pervasive VPN usage as a defense.

11. The Main Event: MongoDB's Security Catastrophe

(140:59 - 167:21)

Script Kiddie Ransomware Boom:

• The "Darknet Army" shares a step-by-step guide to making money by finding open MongoDB instances (often via bad Docker images) with no protection, deleting the data, and leaving a fake ransom note.

  "This isn't some complicated tech heavy process. You don't need to know coding, hacking or anything technical. If you can copy, paste and click, you're good to go." – Darknet Army (142:09)

• The bar is so low, these attacks now target the laziest configuration errors and require no technical skill.

Key Data:

• Flare Systems' analysis (with honeypots):
  • Thousands of public MongoDB servers with default/no authentication (port 27017)
  • 1,416 servers already compromised and wiped, with ransom notes demanding ~$600 in Bitcoin
  • Most ransom demand wallets trace back to a single dominant actor, reinforcing how scalable and impersonal these campaigns are

Root Causes:

• Tutorials, container images, and copy-paste errors spread insecure defaults ("bind to all interfaces, no password")
• Systematic security misconfigurations, not sophisticated exploits, fuel these attacks.

Steve's Takeaways:

1. Authentication should never be the only line of defense; never expose anything requiring login directly to the Internet.

   "Only servers that should ever be exposed... are those meant to be accessed anonymously by everyone." – Steve (162:35)

2. AI will help find and fix more software bugs, but it can't fix operator laziness or organizational sloppiness.

"AI may be getting smarter, it also shows no signs nor hope of being able to make us humans any less dumb." – Steve (165:49)

Notable Quotes & Moments

• On Supply Chain Attacks:

  "It’s one thing for a program to get infected; it’s another for the update mechanism to be subverted." – Steve (24:28)

• On Bug Bounty Spam:

  "We will ban you and ridicule you in public if you waste our time on crap reports." – Curl's new security policy (75:49)

• On AI Security Breakthroughs:

  "AI managed to find 15 out of a total of 16 CVEs in a system of code as carefully composed, maintained and scrutinized as OpenSSL. It's truly a big deal." – Steve (101:22)

• On MongoDB Ransomware:

  "As long as insecure deployment patterns continue... these attacks will remain cheap, scalable and profitable for threat actors, and costly for organizations." – Flare Systems, via Steve (149:50) "It's not that it’s impossible for authentication to work. It's that it absolutely must not be relied upon to work. It should never be the only thing standing between attackers and disaster." – Steve (161:22)

• On Human Error:

  "AI may help us find flaws in our software... But unfortunately... [it] also shows no signs nor hope of being able to make us humans any less dumb." – Steve (165:49)

Timestamps for Important Segments

• AI Taking Over Curation & Coding (06:05 – 08:54)
• Supply Chain Attacks: Notepad++ & Antivirus (23:23 – 50:49)
• Apple’s iOS Location Privacy (55:05 – 62:59)
• Bug Bounty Flood & Curl’s Policy Change (68:09 – 87:38)
• AI Security Breakthrough (OpenSSL) (88:09 – 109:49)
• MongoDB Ransomware Rampage (140:59 – 167:21)

Conclusion & Final Thoughts

This episode underscores two intertwined trends:

• At the bottom: Security disasters (like open MongoDB) are flourishing, precisely because so many organizations copy insecure defaults, don’t lock down public services, and fail to follow even basic best practices.
• At the top: An industrial revolution is underway; AI is supercharging both bug finding and (for bad actors) report spamming, shifting the security industry’s ground beneath our feet.

Steve’s closing message:
AI is poised to eliminate many software bugs, possibly within our lifetimes—but it cannot compensate for a lack of human diligence, prudence, or intelligence. The "Mongo problem" isn’t technical; it’s organizational and human.