There is a known problem in SharePoint (and Teams*) – complicated permissions system. Site owners/administrators provide access, site contributors upload documents and nobody knows – who has access to their sites. As a result – sometimes sensitive documents become overshared (over-exposed).
The biggest concern is when sites content is shared with “Everyone”. How do we find sites shared with “Everyone” in a large Microsoft 365 environment so for this sites we can request permissions review?
(*) Microsoft with the introduction of Teams had to simplify permissions in SharePoint – since there should only be 3 types of access levels – owner, member and visitor. It was… in some ways, but in other ways it made things worse.
Solution #1 (3-rd party tools)
You are lucky if you can use 3-rd party tools (e.g. ShareGate, SysKit Point, AvePoint, Metalogix etc.), with the ability to get full permissions report. Though there might be a problem to get full permissions report for all tenant sites if your m365 environment is not small. Some tools allow you to get tenant-wide permissions report for specific Ids – this option should work better for large environments.
Still there might be another problem. Consider the following. When I say “shared with Everyone” – I actually mean at least 3 possible “everyone” system logins:
- Everyone except external users
- All users
– those are system id’s, but what if there are other ids – e.g. migrated from on-prem or cloud-born custom security or Microsoft 365 groups in tenant that also includes everyone (e.g. dynamic security group that includes all org accounts)? How do you think this group will be identified as “Everyone” group? So – you’d also know which groups include “all” or “almost all” users and get report for these groups also.
Obviously this option #1 is not free, as it requires licenses to be obtained. And still it worth to consider option 3.
Solution #2 (PowerShell “Brute force”)
You can get full permissions report per site or for entire tenant with PowerShell, which if free… The only you need is to write a script yourself or find existing one. Sounds easy?
Well, first problem is it takes a decent amount of time and competences to write such script. If if you find one – it would require some skills to adopt it and run. Frankly say, I have not seen so far scripts that are ready OotB to do that job. And it is not a good idea to run scripts you are not fully confident with against production environment.
Another possible problem – size of environment. The script I designed and use to get comprehensive permissions report might run hours against a good site – if I need full details on site/subsites, lists/libraries, folders and list items levels. So if you have less than 1000 sites – probably this approach can fly. But if your environment is 10K+ sites – it will take forever. So the approach might not work for large enterprise environments.
One might say – we can limit report with root web permissions only to get it faster. But this would not be accurate. And what is not accurate in the IT security – lead to even bigger risks. So, we need check permissions up to every item level deep, as even one file with sensitive info shared inappropriately can cause security issue. (Btw, 3-rd party tools usually by default limit reports to libraries level, so check reporting options)
The other issue with this approach… Let say you got full permissions report… It would look like “resource -> group -> permissions”… How do you know for each group – what is the group in terms of membership?
Ok, if this solution is not easy to get working – what are other options?
Solution #3 (Search-based)
This solution is based on simple but clever idea: why do we need to iterate through all the tenant documents/items if all the content is already crawled by search? Search is also respect permissions. Can we just use search to get files shared with Everyone? Let us see.
What if we use some dummy user account with no specific permissions provided and no group memberships and try to search content on behalf of that account. The idea is if this user can see some data – then these data is open to everyone.
Check this and this articles. Can we get results programmatically (e.g. with PowerShell)? Can we use Microsoft Graph search API? Sure. Check also this article “How to search against SharePoint Online Content with Microsoft Graph search API with PowerShell“.
Solution #3 Option #1 – search through tenant
With this option we would use search query “*” and all 5 possible SharePoint entities – driveItem’,’listItem’,’list’,’drive’,’site’ to find everything that is shared with everyone. We’d pull results with paging (we’d use “from” option in a loop to pull all results). After we get all results – we’d select only unique site collections. But! We might have some problems here.
Problem #1. Again, for small environments or if there are not much “Open” sites – it would work. But for large enterprise environments the problem is the same as in “brute force”. Search would returns too many results – and it might take weeks (exact time is unpredictable) to get all of them. (Surely there are sites “legally” shared with everyone, public Office 365 group based sites, communication sites… So your search will be flooded with content from sites you already know are shared with all).
Problem #2. We are getting results with paging. But recently Microsoft started limiting number of returning results. E.g. your search request result might say like “there are 3659735 total hits” but after result number 1000 it just stops returning anything, even with paging.
Solution#3 Option #2 – loop sites
The idea is: why do we need to get all search results if even one result from a site would be enough to put the site to the list of “open” sites. In other words, we do not need all results from the site, we only need to know if there are any results from the site, at least one – so we know if the site is open for everyone or not.
So, consider the following approach:
- You get list of all sites in tenant.
- You run search request against each site in the loop
(e.g. consider KQL option “Site: https://yourTenant.SharePoint.com/sites/YourSite”.
If at least something found in the site – add the site to the “Open Sites” list.
With this approach you will get list of sites shared with “Everyone…” in a predictable time.
Solution#3 Option #3 – exclude known “open” sites
There are sites “legally” shared with everyone – e.g. corporate portal, department communication sites, public teams, public Viva Engage communities etc. If it is know that these sites are public – you can exclude them from all sites list – so in the “Solution#3 Option #2 – loop sites” – you’d loop only through sites that are not supposed to be public. I know – percentage of “legally public” sites in tenant to all sites is a relatively small number, so should not significantly decrease elapsed time… but still.
Pros and cons of the Solution # 3
Pro: the only fast enough (at least predictable time to complete) and accurate enough to rely on solution.
Con 1 : crawling and indexing takes time, so search-based reports can miss recent changes in data and permissions
Con 2: this approach cannot be automated (since we need an interactive authentication). I.e. we need to run it manually every time.
Con 3: After we get all sites shared with everyone – we do not know – at what level permissions are broken and provided to everyone. It might be entire site or one file. If you want to know what exactly is shared with everyone – run permissions report against this shortlist.
Note 1: consider there are resources like “Styles Library” shared with everyone by default, especially on migrated sites
Note 2: There are might be security groups intended to hold all or part of the enterprise (e.g. “All employee” or “all contractors”). If the enterprise comprises from several businesses or regions – it might be “All Business 1” or “All EMEA”… you got the idea.
You can modify search-based solution if you add your dummy account you are running search on behalf of to some of theses groups to find out if there are resources shared maybe not with everyone but with all “North America based” users or with “all employees”, which might make sense also.
Note 3: separate service, but consider implementing/using sensitivity labels. At least you can start with high-sensitive sites. With sensitivity labels – site owners/member would know – what kind of site they are working on.
Ok, we know list of SharePoint resources shared with everyone, but what would be the next step? Should we communicate to site owners – if so how to let site owners know that there are resources shared with Everyone… on their sites.
To be continued…
- Full Report on SharePoint Online Site Access and Permissions
- Calling MS Graph Search API against SharePoint content