
The easiest way to prevent technology chaos is carefully managing which tools and services can access your users and data. Evaluating new edTech tools is a critical part of IT security, and a tremendously tedious process. 😵💫
I needed a better way to quickly evaluate new tools for the districts I work with, so I turned to Gemini and created a custom Gem to analyze privacy policies based on a custom set of criteria.

This Gem acts as an expert legal assistant, analyzing privacy policies against my required criteria:
- Clearly articulated privacy policy
- Student data is not used for advertising or marketing
- Data is not sold or shared with third parties
- Presence of data security safeguards
- Regulatory compliance with SIPA, COPPA, FERPA
- Identify any unique policies for users under the age of 18
- For AI tools, determine if data is retained for model training and improvements
So far, I’ve been pleased with the effectiveness of this custom Gem. I used it to evaluate a couple of privacy policies that I knew pretty well (Kami and Google workspace), and both platforms passed with ease.
Related Post: Understanding Gemini data privacy for EDU domains
I also tested the Gem on ChatGPTs public policy and the Gem indicated that OpenAI does not comply with my required criteria, just as I anticipated.

Try it out!
Earlier this month, Google launched Gem sharing which means that I can share my custom Gem with you! Fill out the form below to give my custom gem a try.
Copy and Customize
If you like the idea of using AI to evaluate privacy policies, it is best practice to make a copy of my original so that you can add your own criteria and modifications. This will also ensure that any future changes I make to my Gem won’t impact you.
Here are some additional criteria you might want to add:
- Local or state data compliance rules
- If you have a district evaluation rubric, attach it to the Gem for more personalized recommendations.
- A list of district requirements (such as “supports Clever integration” or “syncs with Powerschool”)




Reader Interactions