Recommended Posts

My main router is an old (and I mean REALLY old) Netgear RP614 v2, and it is not vulnerable. :huh:

Edit: Happy Birthday, Budman!!!! :punk: :pint: :pint: :pint: :pint: :pint: :pint: :pint: :pint: :pint: :pint: :pint: :pint:

To those who don't get it and want the short version:

The problem is that some routers will respond to UPnP requests, wherever they're coming from. If they're coming from the LAN--no problem (unless you don't trust other machines/devices within your own LAN). If they're coming from the WAN port--then that's bad and you should disable it.

If you need more details than that, then listen to the podcast on the GRC site.

I disable it anyway. The fact that UPnP, by design, lets any application communicate with the router and open ports should make any security conscious user uneasy.

Not really. At this point you have already lost and been invaded anyway, and the route out should be of much more concern than than the route in, and if the program in question can open a route in, it's also capable to two way communication without opening a port.

that site is scaremongering at best anyway. notice how it ONLY reports how many "open" routers has been found with the test, not how many secure ones.

It would allow any malicious program to actively contact your router, open whatever ports it wants, and then transmit data through those ports all without your knowledge.... pretty big security hole if you ask me.

From the inside, at which point you've already lost and UPnP isn't needed anyway

THE EQUIPMENT AT THE TARGET IP ADDRESS

DID NOT RESPOND TO OUR UPnP PROBES!

Amped Wireless R20000G Passed! UPnP Enabled

Also mine's not on the affected devices list either!!!! Go Amped!!!!

Also mine shipped with uPnP disabled as well as WPS disabled. Extra points for them!

So is this a just rubbish. Default settings on router since I bought it and I got a pass.

I just have a forward to my web server.

Router is

Netgear DG834GT with the firmware updated to the latest.

I have an Airport Extreme router. I don't see an option for UPnP on the Airport Utility. Then again, it doesn't have many options at all.

THE EQUIPMENT AT THE TARGET IP ADDRESS

SUED OUR UPnP PROBES!

Today I had to help setup a computer for a little old lady. While I was there I ran the UPnP test. Her's failed! She had a D-link (Go figure) .... logged into the router and turned off UPnP, ran the test again and then it passed. So the rest does work!

Still not sure how people think this is suddenly new... it's been like that for a while. Didn't one of the US agencies mention this years ago?

I've always suggested to disable UPnP.

Because UPnP should NEVER be on the WAN side (internet). This means a bad guy could send a packet to your IP and if your router responds (Which is what this test is for) he could open a port in your route from the outside (Internet)

pfSense, enough said.

Not practical for the average consumer, enough said.

From the inside, at which point you've already lost and UPnP isn't needed anyway

Not from the inside, the exploit is that it responds to UPnP from the WAN side, that's the problem.

Not really. At this point you have already lost and been invaded anyway, and the route out should be of much more concern than than the route in, and if the program in question can open a route in, it's also capable to two way communication without opening a port.

that site is scaremongering at best anyway. notice how it ONLY reports how many "open" routers has been found with the test, not how many secure ones.

I think your failing to understand the exploit, typically the packet is formed on the LAN side from an application, which is passed to the router, the router opens up the ports requested. The problem is here, if you are running one of the exploitable routers, ANYONE from the WAN side, can sent a correctly formed packet to your router, over the net, and your router will open the port for them. This should never be allowed on the WAN interface.

that site is scaremongering at best anyway. notice how it ONLY reports how many "open" routers has been found with the test, not how many secure ones.

That is because MOST routers SHOULD pass the test!!! There shouldn't be very many routers that by default have UPnP on the WAN. The people who have run this test in this thread have proven that.

It's a MUCH bigger deal if you fail the test than if you pass it.

uPnP is the dumbest idea. whats the point of the firewall if applications are just going to open dat dere ports anyways? if you get a piece of malware that runs a server on your pc,it will just open the ports it wants,and runs beautifully. if you open your own ports,you at least know what you're getting yourself into. you don't even have to have malware. you might have a vulnerable application that is actively listening on a port.

guys please be sure you specify the router you are using for the tests... some of you didn't and that's not helpful...

That is because MOST routers SHOULD pass the test!!! There shouldn't be very many routers that by default have UPnP on the WAN. The people who have run this test in this thread have proven that.

It's a MUCH bigger deal if you fail the test than if you pass it.

My Amped Wireless R20000G and my R10000 both shipped with UPNP disabled.

I enabled on both and they pass the test and "do not respond"

guys please be sure you specify the router you are using for the tests... some of you didn't and that's not helpful...

My Amped Wireless R20000G and my R10000 both shipped with UPNP disabled.

I enabled on both and they pass the test and "do not respond"

I agree they could list their router. But we are already more than 4 pages in. So people could either flip through the pages looking to see if someone who ran the test has the same router than them, or they could just go to the site and click the button.

This thread was created not really as a list of routers affected but as away people can test themselves against the issue.

uPnP is the dumbest idea. whats the point of the firewall if applications are just going to open dat dere ports anyways? if you get a piece of malware that runs a server on your pc,it will just open the ports it wants,and runs beautifully. if you open your own ports,you at least know what you're getting yourself into. you don't even have to have malware. you might have a vulnerable application that is actively listening on a port.

It's so that when somebodies mother who views the computer as a magic box wants to make a Skype call with somebody, she doesn't have to reconfigure the firewall to let things pass through.

I'm running a dual stack (v4/v6) setup, and the UPnP daemon I'm running doesn't support the v6 side yet so any open ports only happen for v4 traffic. It's surprisingly annoying to track down what uses what ports to add them to the firewall.

This topic is now closed to further replies.
  • Recently Browsing   0 members

    • No registered users viewing this page.
  • Posts

    • Hi all. I have a seven year old Acer laptop which according to Acer website can't receive the new MS updates. This device lives in my bedroom and is purely a consumption device. No gaming, no nothing. I have followed online instructions and turned off secure boot which now gives me a green tick in W11 Security. Now, I have MWB Premium now for several years and I'm reasonably careful with my online activity with this laptop. I'm also not a gamer in any way so I'm not asking much from this laptop. Given that, am I safe to continue using this device given I've turned off secure boot? That's it, that's my question.
    • finally [Taskbar] Taskbar customization just got easier. As we continue to make improvements to the Taskbar experience mentioned last month, we've introduced a dedicated Taskbar Size setting, making it simpler to find, understand, and personalize your ideal taskbar experience.
    • Let me get this straight... It was a web interface for Gmail, so if privacy at Google wasn't concerning enough you'd be going through two companies. And their big feature was the very thing that would make people consider dumping Gmail.
    • Microsoft's fast coding model MAI-Code-1-Flash comes to Copilot Business and Enterprise by Karthik Mudaliar Microsoft’s recently announced MAI-Code-1-Flash model is now generally available to GitHub Copilot Business and Copilot Enterprise customers. With this support, organizations can have more centralized policy controls and billing while finally being able to use Microsoft’s lightweight, first-party coding model. According to GitHub’s announcement, Business and Enterprise plan administrators must enable the MAI-Code-1-Flash policy in Copilot settings before developers can access the model. Microsoft says that MAI-Code-1-Flash is for fast, iterative coding work rather than the most demanding architectural or debugging tasks. GitHub’s official model comparison page says that the model is great for "general-purpose coding and writing," while it excels at fast, accurate code completions and explanations Microsoft introduced MAI-Code-1-Flash on June 2 as part of a broader collection of internally developed MAI models. GitHub subsequently expanded support to Copilot CLI, the Copilot cloud agent, GitHub.com chat, GitHub Mobile, Visual Studio, JetBrains IDEs, Eclipse, and Xcode, but said support for managed Business and Enterprise customers was still on the way. In Microsoft’s own benchmark testing, MAI-Code-1-Flash scored 51.2% on SWE-Bench Pro, compared with 35.2% for Anthropic’s Claude Haiku 4.5. Microsoft also claimed that the model used up to 60% fewer tokens on SWE-Bench Verified. Do note that these are vendor-run results rather than independent measurements. The model is billed at provider list pricing under GitHub’s usage-based system. GitHub currently lists MAI-Code-1-Flash at $0.75 per million input tokens, $0.075 per million cached input tokens, and $4.50 per million output tokens. For organizations, the main incentive to use MAI-Code-1-Flash is likely to be efficiency rather than maximum capability. A smaller model that responds quickly and limits unnecessary output is quite useful for repetitive agent tasks at scale, especially after GitHub Copilot’s move toward usage-based billing. The "Flash" model is recommended for fast work and not necessarily for huge repositories with loads of context. It's better if teams compare their output with other larger models, especially if they're working on security-sensitive changes and complex, multi-file work.
  • Recent Achievements

    • One Year In
      bernmeister earned a badge
      One Year In
    • Week One Done
      Scoobystu earned a badge
      Week One Done
    • Week One Done
      tuben earned a badge
      Week One Done
    • First Post
      OffsetAbs earned a badge
      First Post
    • Reacting Well
      OffsetAbs earned a badge
      Reacting Well
  • Popular Contributors

    1. 1
      +primortal
      462
    2. 2
      +Edouard
      213
    3. 3
      PsYcHoKiLLa
      157
    4. 4
      Steven P.
      73
    5. 5
      FloatingFatMan
      71
  • Tell a friend

    Love Neowin? Tell a friend!