Recommended Posts

You make them a normal user on the computer. That is how you do it. If they use a boot disk to get around it, set a bios password and put security screws on the case so they can't open it. If they continue to break policy it is grounds for termination or removal of computer rights (at least that should be in your policy).

Power users does not allow you to add admins to the computer. You would need to be an admin or a group that has local admin rights. Or a boot disk like the free hirens disk that you can download and boot off of.

"We had user today add themselves as a local admin,"

As sc302 mentions "boot disk to get around it, set a bios password"

If I have physical access to the box - I can just boot one of many different tools to change the local admin account password. Log in with that for what I need, or log in with that and then give whatever other account I want local admin as well.

You can not prevent that from happening with a gpo.. The box would have to be setup with a bios password to prevent booting from removable media, be it cd or usb, etc. And you also need to prevent pxe - or I could just boot the tool I need to change the local admin password via pxe if so desired, etc.

  • Like 2

Windows 2008 domain with domain controllers running '08 R2.

We had user today add themselves as a local admin, giving himself full rights to that machine on a Windows 7 machine.

How can I prevent this with a GPO?

Group Policy Preferences. This will properly layer over multiple GPO's targeting the same group.

Computer Configuration\Preferences\Control Panel Settings\Local Users and Groups.

Create a "New Local Group" and from the drop down caret under Group name: select "Administrators (built-in)

Checkmark "Delete all member users"

Checkmark "Delete all member groups"

Now select the "Add" and then "..." buttons to query for domain groups or user objects. Do not type in the Name: field manually unless you are defining a local computer user or group object that will still be a member, otherwise you may not properly attach the domain object SID to the GPP. You may wish to add the local computer "Administrator (built-in)" user to this group.

Under the Common tab you should select "Remove this item when it?s no longer applied." and select "No" so that all members added to the group are removed when the GPO no longer is used against the computer. Beware that if you do not have a higher level GPO that automatically adds a local Administrator to this group that it is possible to remove all Administrators from a computer.

Beware on using "Delete all local users" against servers. You will discover scenarios where your GPO will cease to function on servers containing certain roles (I believe it involves the Configuration Manager agent being present), and then you may have almost everyone trapped out of the server until you create a lower level GPO that fixes the issue and wait for your GPO refresh timer to execute. In this particular case if you need to control local users, have a parent GPO control removal, and a layered GPO control the actual members. It'll scream at you in the event logs when Delete all local users fails to function.

If you do this, even if someone adds an account to the Administrators group, your next GPO refresh will undo the change. If someone adds a user to the group while the machine is offline, the GPO refresh at startup will likely remove the user from the group before or during logon. People who know how to launch PowerShell under the System context of the Logon screen however will be able to defeat this GPO. If you really want to prevent offline attacks, use BitLocker, but BitLocker will not defend against an online attack that enables a CLI to work at the logon screen.

As I recall, this GPP has no effect against built-in groups on Domain Controllers.

  • Like 3

Group Policy Preferences. This will properly layer over multiple GPO's targeting the same group.

Computer Configuration\Preferences\Control Panel Settings\Local Users and Groups.

Create a "New Local Group" and from the drop down caret under Group name: select "Administrators (built-in)

Checkmark "Delete all member users"

Checkmark "Delete all member groups"

Now select the "Add" and then "..." buttons to query for domain groups or user objects. Do not type in the Name: field manually unless you are defining a local computer user or group object that will still be a member, otherwise you may not properly attach the domain object SID to the GPP. You may wish to add the local computer "Administrator (built-in)" user to this group.

Under the Common tab you should select "Remove this item when it?s no longer applied." and select "No" so that all members added to the group are removed when the GPO no longer is used against the computer. Beware that if you do not have a higher level GPO that automatically adds a local Administrator to this group that it is possible to remove all Administrators from a computer.

Beware on using "Delete all local users" against servers. You will discover scenarios where your GPO will cease to function on servers containing certain roles (I believe it involves the Configuration Manager agent being present), and then you may have almost everyone trapped out of the server until you create a lower level GPO that fixes the issue and wait for your GPO refresh timer to execute. In this particular case if you need to control local users, have a parent GPO control removal, and a layered GPO control the actual members. It'll scream at you in the event logs when Delete all local users fails to function.

If you do this, even if someone adds an account to the Administrators group, your next GPO refresh will undo the change.

As I recall, this GPP has no effect against built-in groups on Domain Controllers.

The damage is already done at that point. There is no gpo that prevents this.

  • Like 2

The damage is already done at that point. There is no gpo that prevents this.

Depends what the actual story is, which the OP barely gave details on. What I described helps mitigate. (BTW, you responded to my original message prior to a bit of editing)

To prevent offline attacks, the only real "solution" is to manage the machines with BitLocker, a TPM, and Network Unlock.

A BIOS System password is only effect against "some" computers with properly designed firmware. A large majority that I've encountered do not block the F12 (or equivalent) firmware/BIOS boot menus even if a System password is present, including some of Dell's business line machines. Only some actually require authentication if a system password is present. I have some Precision workstations that do intrusion detection great, but only a BIOS user password will prevent a user from calling on the boot menu (and of course block them from using the computer at all without support). I don't believe any vendor is 100% consistent across their motherboard models when it comes to securing its BIOS/Firmware boot menu.

Also, when properly managed, "BitLocker+TPM+Network Unlock" is the better solution than any firmware block or physical lockdown because it requires the end user actually have technical skills. They need to have successful online attacks before an offline attack becomes possible. At this point most failures will be the result of desktop mismanagement.

Obviously it?s a bit trickier on mobile systems, as Network Unlock likely becomes impossible and you have to replace it with +PIN/+USB.

Sorry for the brevity in the OP. I hammered out the question before leaving the office for the day. I probably should have waited until I had all the details before posting. An area tech called me with the problem and the details were vague.

It is my understanding that the user launched the user account applet and made the changes. I was hoping that there was a GPO that disabled access to that particular applet.?.

Thanks for the responses.

When I get to work tomorrow morning I'll set up a test user with the same privilages and try it out and see if I can figure out how he did what he did.

It is my understanding that the user launched the user account applet and made the changes. I was hoping that there was a GPO that disabled access to that particular applet.?.

Then the problem lies with your lack of basic security then, nothing GP will fix - you've set his account up as an admin.

Agree with grounds for termination and so on, but also worth pointing out, try and find out what possessed the user to get Admin rights. Did he do it because he is an ass who wants to install some dodgy Facebook games, or is there something wrong with his computer that hasn't been addressed and he was trying to take matters into his own hands to fix it out of frustration?

Kaedrin -- are you talking about restricted groups? Seems like a really long explanation of restricted groups to me. Which sure you can restrict who is in your admins group. And yes that is a great idea and normally an audit requirement anyway..

But if I have local admin - its real easy to block gpo being pushed from the domain..

Now sure if you want to go the encrypted route - this can also prevent the boot tools to change the admin account. But normally your not trying to keep out the elite hackers here.. Your keeping billy joe bob from running some boot tool he found on the net, etc.

But yeah if all he did was launch user manager -- then he had rights in the first place. Does someone have domain users in the Domain Admins group ;)

Kaedrin -- are you talking about restricted groups?

Restricted Groups is legacy. They have extremely limited functionality compared to GPP Groups, and as I recall cannot be layered across multiple GPO's. Unless the target is a pre-Vista system, GPP Groups should be used instead. I abandoned Restricted Groups entirely once Vista SP1 & 2008 SP1 were released.

Even if you set a boot bios password to prevent booting from CD the passwords can usually be reset. However if you enforece bitlocker with the key being backed (with hardware TPM) up to AD and only recoverable from AD admins there is no way they can use any off the street tool to add thierselves as Admin. First they would have to have access to the AD and have rights to view the key. Note Admins can still boot and use the recovery MS DART toolsets and reset passwords or whatever with the recovery key. There is no way the user will be able to boot from cd and give himself admin rights.

There is a small chance of privilege level escalations using say a faulty cisco vpn client allowing the users to get system access and then give themselves root from a running machine. Also there is a small chance they could freeze the memory (actualy temp wise) and read the bitlocker key from an additional machine. However the general user with all his "hacker" tools arent' going to bypass a full system encryption (assuming you have a TPM module in place)

You don't say if you do have a TPM enabled machines or not.

However s/he already OWNS this machine. The only way to ensure they don't give themselves rights again is to re-image. As as long as he or she had admin rights to begin with, they may of installed a system level back door that simply gives the right back even after you removed them from the Admin group.

Do you make them sign an user agreement, or if they do have elevated rights a privledged level access agreement?

A little more detail;

This is at a school district and the kid in quesiton is a seventh grader. He renamed the local admin account and changed its password. Those student accounts are not local admins on the computers. After testing it with an account that we copied from his account, the only way he could have done this is with a bootable device, be it a cd, dvd, or usb drive. Thankfully with the security in place, being a local admin doesn't give him any rights on the network or domain. About the only thing he can do differently is install software locally. After going through his internet search history, he is looking for remote viewing software, specifically TeamViewer. He did not install any software though - maybe the bell rang and he ran out of time?

We're hatching a plan to catch him doing it so that we can nail down his process and figure out how much he knows. Maybe a key logger, or remote viewing software - like VNC. He could have used a product like Ophcrack and now he knows the local admin password on all the campus computers.

In the future we'll set the bios to only boot to the SATA drive and set a password on the bios to prevent changes being made to the boot order.

Any other thoughts/recommendations?

to catch a thief...

http://www.spectorsoft.com

Edit: also on some bios's you can set what devices are allowed to boot from not just boot priority...hitting the option to boot of a different device will yield only what devices you choose (ie, the hard drive).

The other issue is that it is pretty easy to reset this if you can pull the cover off when no one is looking, so putting in a security screw to prevent cover removal without special tools may also help.

If your admin password is simple enough to be cracked locally by Ophcrack in a reasonable amount of time, that's a huge problem. Presumably you have a descent password policy and that is not the case, so I would rule out the student having access to the local admin password for all the machines.

Also, have you heard of Kon-Boot before? Its an extremely simple (and effective) method for gaining access to virtually any local Windows account, including the admin account. It would allow someone to boot from a disc, login as admin, then change the admin password. One could also login to any domain account that has been established on the local computer (i.e. a domain admin, teacher, or other student who logged in and created a profile on that computer), but since Kon-Boot bypasses the password instead of cracking it, the user won't have access to any domain resources.

Depends what the actual story is, which the OP barely gave details on. What I described helps mitigate. (BTW, you responded to my original message prior to a bit of editing)

To prevent offline attacks, the only real "solution" is to manage the machines with BitLocker, a TPM, and Network Unlock.

A BIOS System password is only effect against "some" computers with properly designed firmware. A large majority that I've encountered do not block the F12 (or equivalent) firmware/BIOS boot menus even if a System password is present, including some of Dell's business line machines. Only some actually require authentication if a system password is present. I have some Precision workstations that do intrusion detection great, but only a BIOS user password will prevent a user from calling on the boot menu (and of course block them from using the computer at all without support). I don't believe any vendor is 100% consistent across their motherboard models when it comes to securing its BIOS/Firmware boot menu.

Also, when properly managed, "BitLocker+TPM+Network Unlock" is the better solution than any firmware block or physical lockdown because it requires the end user actually have technical skills. They need to have successful online attacks before an offline attack becomes possible. At this point most failures will be the result of desktop mismanagement.

Obviously it?s a bit trickier on mobile systems, as Network Unlock likely becomes impossible and you have to replace it with +PIN/+USB.

A firmware password would be used to stop the booting off of cd/dvd so they cant boot up a password change dvd

"he knows the local admin password on all the campus computers."

While I agree the password should be of significant strength to prevent bruteforce/rainbow tables, etc. You should also not have the same password on every device. They should all be different.

I use to just run a script to change the local admin password on every single machine in the domain. It was a random 20 digit that just dumped out from online password creator - http://www.pctools.com/guides/password/ I just pop these into my script and store in our password log.. Rare that was ever needed, you normally just used your own account that had local rights on all the machines, etc. Only in the case that boxes trust got messed up with the domain or something would you have to use local admin.

This also allowed us to give out local admin in the worse case scenario user was offline and something when wrong where you had to walk him through something with admin rights.. All you gave him was his box, and next time he his box was on the domain you would change it, etc.

BudMan, that's a really cool idea! None of the companies I have worked for have done that, but they probably should have. That seems like a much better method than generating a strong local admin password that must be changed once every three months. Everyone in IT basically uses their domain admin privileges anyway.

we had to change them every 3 months as well ;) Which is why the script!! It unrealistic to think you could change password if you had to touch every machine.. And your going to typo it for sure as well ;)

Might not have been that bad on the 50 some local servers, but not going to do it on the 900 plus user boxes.. And what about laptops that rarely come into the office etc.. Just needed to catch them while they are online via the vpn and 2 seconds there you go their local admin is changed.

Oh they were secure as well -- other guys use to bitch if they ever had to use them, or even the domain admin account which again used random 20.. As you said everyone normally used their personal accounts that had the permissions they needed. But every now and then you would have to break out the domain admin account password from the safe. and **** like this can be a pain to type in #luc&ouk?aqL6#iEwr+e

which is why the phonetics come in real handy ;)

(Hash - lima - uniform - charlie - Ampersand - oscar - uniform - kilo - Question - alpha - quebec - LIMA - Six - Hash - india - ECHO - whiskey - romeo - Plus - echo)

A firmware password would be used to stop the booting off of cd/dvd so they cant boot up a password change dvd

As I said, there is no guarantee a motherboard actually secures the boot menu just because you set a firmware password. There is no consistent behavior across motherboards even from Dell. If you have 6+ generations worth of different computers, I?m sure that less than half of them will actually behave the same way. Some will ask for authentication, and some won?t care that there is a password and will simply always allow the boot menu to function.

Give an example of something that allows you set a firmware password but does not allow you to lock down the boot, or still allows access to the boot menu.

Remove everything but the hdd from boot option, turn off feature to access boot menu, set password.

Even if the boot menu is available, if you alter the boot options then set a password on the firmware you should be good. I don't recall ever in the 30+ years in working/playing with a computers -- once they starting adding bios passwords ;) Not having the ability to lock out what the computer could boot from.

Shoot most users can not figure out how to change the boot order when firmware is NOT locked ;) Or for that matter press F# the system even flashes on the screen that says to change boot order.. So just the fact of changing the boot order to not boot cd/dvd/usb/pxe before the hard should keep most users from being able to run the boot tools.

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

    • No registered users viewing this page.
  • Posts

    • Just pull a 4Chan and ignore the UK gov, or better troll them. It's not like they can enforce the fine across border.
    • It has NEVER been shown that all these overreaching creepy methods of surveillance have ever saved a child or prevented a terrorist attack. Not a single one. It's the kind of people like you who just wave it away as "paranoid conspiracy" that makes big tech and governments this creepy mass data hoarding entities. Not only that, 3/4 of these surveillance ideas undermine the very foundations of safe online communication because they always want to have a backdoor in everything "just in case" they might need it to... checks the notes "save the children". If you put a backdoor into encryption chain there is no encryption chain anymore. You know what encryption keeps safe? Your medical records, your online shopping and credit card during payment, your photos in the cloud, your emails, your passwords, everything. There is ZERO guarantee only the good guys will use it. And if you think police suddenly can't apprehend child abusers because of encryption, Epstein was running his entire sex trafficking ring using GMail which is not even encrypted end to end. Or to make matters even worse, USA has a **** and a good buddy of Epstein as a president. Absolutely NOTHING has been done to address it. Maxwell just got a better "hotel" room as a reward. This clearly shows how they absolutely don't really care about the children but they care about the absolute control over all of us. And you're defending them here. Good grief. On top of constant attempts to insert backdoors into encryption chain, the entire age verification nonsense is again entirely over reaching, creepy, invades everyone's privacy with premise of yet again "protecting the children" instead of demanding device makers to provide simple and powerful tools for PARENTS to control how their children use devices and what they do on them. THIS would be the way, not the stupid age verification for everyone. Imagine if government would be dictating companies how their phones work and not the company's IT department. The parents should be the IT department to their children. And for everyone excusing "they are not knowledgeable enough" buuuuuulsheat. We live in a digital age, if you have children now, you absolutely are well versed in digital everything at least to basic extent. If you're not, how do you even function in these times then? Reality is that parents are just lazy and don't want to deal with this. They want government to raise their kids because they are too busy scrolling stupid Instagram and Tiktok or some bs.
    • You could make the argument that K should not be included, but FC, the fried chicken, is not the framework, it's the product. It's the Paint in Paint.NET. A closer analogy is if KFC included the name of the deep fryer they used. HennyPennyFC.
    • Flying as the central point eh... As a massive Spyro fan who has replayed the Reignited Trilogy three times and the originals 4 times... I have some doubts, but maybe...
    • Apple is expanding Private Cloud Compute beyond its own data centers by Pradeep Viswanathan At WWDC 2026, as part of the improved Apple Intelligence capabilities, Apple today announced that it is expanding Private Cloud Compute (PCC), its privacy-focused cloud infrastructure for Apple Intelligence, beyond its own data centers for the first time. Private Cloud Compute was designed to handle Apple Intelligence requests that are too complex to run fully on-device. The PCC system does not store user data and does not allow Apple or anyone else to access user requests. Last year, Apple also expanded its Security Bounty program with rewards of up to $1 million for researchers who could find serious vulnerabilities in PCC. Until now, Apple's PCC data centers were using Apple's own silicon. As part of the expansion, Apple is working with Google and NVIDIA to run new Apple Intelligence workloads on Google Cloud systems powered by NVIDIA GPUs. Apple will be using this new infrastructure to execute more demanding AI tasks while maintaining the same privacy and security guarantees of PCC. The new implementation uses NVIDIA Confidential Computing with NVIDIA GPUs, Intel CPUs with TDX, and Google’s Titan chip. Apple says it has worked with Google to build additional protections beyond a traditional confidential computing deployment. Despite the expansion to third-party data centers, Apple claims that its core PCC requirements remain unchanged, including stateless computation, no privileged runtime access, non-targetability, and verifiable transparency. The company highlighted that it will continue to control the PCC software stack, and Apple devices will only trust PCC software that has been cryptographically approved by Apple. To take security to the next level, Apple mentioned that it is maintaining an append-only ledger of Google Cloud hardware that is part of the PCC fleet. The company claims this will help reduce the risk of supply chain attacks. In addition to AI infrastructure, Apple also worked with Google to use technologies behind the Gemini family of models to build the next generation of Apple Foundation Models to power Apple Intelligence features across on-device and cloud workloads. As expected, for more demanding AI tasks like agentic tool use and complex reasoning, Apple will rely on the expanded PCC infrastructure running on Google Cloud. The expansion of PCC on Google Cloud will gradually ramp toward the full set of protections during the summer preview period. As before, Apple will also publish binaries for public inspection, provide research tooling, and give researchers access to live PCC nodes in research mode through the Apple Security Bounty Program.
  • Recent Achievements

    • Very Popular
      Captain_Eric earned a badge
      Very Popular
    • One Month Later
      amusc earned a badge
      One Month Later
    • One Month Later
      DJC50PLUS earned a badge
      One Month Later
    • Week One Done
      DJC50PLUS earned a badge
      Week One Done
    • Proficient
      Eric Biran went up a rank
      Proficient
  • Popular Contributors

    1. 1
      +primortal
      506
    2. 2
      PsYcHoKiLLa
      222
    3. 3
      ATLien_0
      92
    4. 4
      +Edouard
      86
    5. 5
      Steven P.
      81
  • Tell a friend

    Love Neowin? Tell a friend!