staff member wanting ai in the school

[xicor]

Has anyone deployed AI tools for staff across their school?
 

We currently have Microsoft Copilot, but staff are increasingly asking for something closer to the ChatGPT like being able to easily group chats into projects, keep related work together, and generate images etc

 

The problem I’m finding is that ChatGPT becomes very expensive once you move into the versions with proper management and security controls. ChatGPT Business/Teams seems more affordable, but from an IT admin point of view the controls still feel quite limited. For example, I can't seem to centrally/bulk disable plugins or properly lock down some of the features I wouldn't want staff using.

 

I'm interested to know what others have deployed for staff if any ? or if you using Copilot, ChatGPT, Gemini, Claude, or something you've built yourselves using an API?

 

[3s-gtech]

Copilot only - laid out in our AI policy. This is for Data Protection and budgeting reasons, and also for things like SARs - we need data in places that we can audit and manage. If they really want to use another AI service, they'll need to lay out the business case and resolve these questions.

[xicor]

39 minutes ago, 3s-gtech said:

Copilot only - laid out in our AI policy. This is for Data Protection and budgeting reasons, and also for things like SARs - we need data in places that we can audit and manage. If they really want to use another AI service, they'll need to lay out the business case and resolve these questions.

did you have to pay for extra data controls for copilot or is this the standard license included A1 ?

[DalekSec]

Not a school but:

CoPilot for general use, free tier (that comes with out licesning)

Claude for heavy users, paid for business tier 

[3s-gtech]

15 minutes ago, xicor said:

did you have to pay for extra data controls for copilot or is this the standard license included A1 ?

A1/A3. I spend enough on MS licensing yearly; paying extra for this guff is money in the wind.

[andy_b]

Struggling with this a little.  Chatgpt is demanded as people are unhappy with Copilot's output.

[xicor]

8 minutes ago, andy_b said:

Struggling with this a little.  Chatgpt is demanded as people are unhappy with Copilot's output.

yeah this was another complaint along with lack of project and chat grouping

[ITGuyNW]

Just say yeah we can have chatgpt but costs will be shared with departments..

[nicholab]

Has anyone considered a product like Netwix DLP?  As that could control anywhere that data is going to.

[sigma]

So how many schools are still using free or other non-enterprise versions or have not enabled ZDR policies on the enterprise version? 

 

https://www.wired.com/story/how-to-use-ai-with-your-privacy-intact/

“When you start typing into one of the big three AI chatbots—ChatGPT, Claude, or Gemini—it’s safest to start with a baseline expectation of approximately zero real privacy from anyone who is determined to access your conversation records and has a legal path to obtaining them. That includes the owner of the service, advertisers or other companies they partner with, contractors who help fine-tune the systems, law enforcement agencies, or even someone who manages to subpoena the records as part of a civil lawsuit.

The simplest, strong exception to that rule is a contract between you—or more likely, your employer—and an AI provider that legally prevents them from retaining those records, a provision that’s come to be known as zero data retention, or ZDR. OpenAI, Anthropic, and Google all offer ZDR policies for their enterprise versions, which when enabled generally require that they immediately delete records of users’ interactions with a chatbot as soon as they’re processed.

Even these ZDR policies, which are only available for paid enterprise and developer accounts rather than for average users, have significant exceptions. Anthropic, for instance, doesn’t offer ZDR for its most sophisticated “Mythos-class” models like Fable 5.1, due to what it describes as the potential for misuse like scamming or hacking and even “autonomous misbehavior,” such as AI agents independently hacking targets to carry out their unwitting users’ requests, as has occurred in several high-profile AI-driven breaches.

OpenAI also announced last month that its ZDR implementations will analyze user activity prior to deletion—on the customer’s systems rather than its own, it says—and, if it detects abuse, flag it for the customer organization. OpenAI says its system, known as Private Safety Processing, will even alert OpenAI’s staff if it detects abuse in some cases, though without revealing the content of the conversation to them. Google, too, warns that it logs some Gemini prompts for abuse monitoring even with ZDR enabled, but without the user’s Google ID or IP address included—though that “sanitized” data can in some cases still pinpoint a particular person when the request itself includes identifiable information.”

[jmak]

Also not a school, but I'd advocate getting away from staff signing up to free accounts and using them for work. The school should have a policy prohibiting it - and regularly remind staff. You all know this already, but for the record school days should only be entered on platforms whose privacy policies you've vetted. 

 

I'm possibly being unfair, but my experience of Copilot was so poor compared to any of the free products, it was causing half the problems by driving users to free ChatGPT accounts. Check what you can get for free or low cost from Google as a school. Gemini is pretty decent and Gemini Notebook (previously Notebook LM) is really good. We decided Claude was enough of an improvement that we set up a Team subscription and users who can justify it to their manager will be given a seat. It's paid for by the users cost centre - it's not (yet) part of basic IT provision.

 

I think the difference between the platforms for day to day activities is small, although I prefer the style of writing Claude uses in responses and draft documents (technical documentation I only care about content and clarity; I wouldn't use the output unedited for documents to be shared widely), but we selected it on the preference of our coders. They ran various tests and at the time we selected it, it was significantly ahead of the others. I think now it's closer.