AI in schools: What actually matters before the July 1 deadline

AI is moving fast, and Ohio schools now have a July 1 deadline to have a policy in place.

So what should districts actually be doing right now?

In this Legal Ledger Sidebar, OSBA policy consultant Gamy Narvaez breaks down the steps schools can take now, what matters most in an AI policy and the biggest legal risks to watch.

If your district hasn’t started yet, this is a good place to begin.

00:00 The July 1 AI deadline
00:45 Where districts should start
01:50 What a good AI policy includes
03:00 Start from scratch or use a model?
03:45 FERPA and data privacy risks
05:00 Chatbots and student safety concerns
06:50 AI misuse and deepfakes
08:20 Biggest legal risk right now
09:00 What to do this week

For more resources, training and advocacy updates from the Ohio School Boards Association, visit ohioschoolboards.org.

Transcript

SPEAKER_01 0:00

The following program is a presentation of the Ohio School Boards Association. AI is moving fast, and for Ohio schools, there's now a real deadline on the table. Hey everyone, it's Scott Gerfin. School districts are expected to have an AI policy in place by July 1st. But what does that actually mean? Where do you start? And what should you be paying attention to right now? Well, today we're joined by OSBA policy consultant Gami Navarez to break it all down. Gami, let's get right into it. When you talk to schools, what are you telling them right now? What should they actually be doing?

SPEAKER_00 0:44

Yeah, I mean, well, we're definitely telling them about that timeline, that's for sure. Um, so although House Bill 96 requires a policy addressing appropriate AI uses for staff and students, it doesn't actually include any specific requirements. So I think there's a gap that these districts need to fill, and we would say the the first step is to develop an AI framework to help govern their use. So maybe a few practical steps that I would leave any district is first of all to adopt an AI policy as soon as possible. Um and uh use that policy to guide your district and align AI use. So whether that's ODU's model, um, our model uh for OSBA, which is policy EDEC, if you use our policy, or some custom model, um, then we would say it's worth taking an inventory of existing AI tools that are being used in your schools already by staff and students, which you may not even know uh kind of what tools are already being used, and establish some sort of vetting and approval process for all of those AI tools.

SPEAKER_01 1:52

Yeah. So when a district sits down to write this policy, uh what's most important to get right? Not the legal jargon, uh, but the stuff that actually matters uh day to day.

SPEAKER_00 2:06

Well, since we already wrote a policy, I'll use that kind of as a framework. I would say that uh a good district policy does need to clearly lay out obviously what is allowed and then what's prohibited as far as AI use, but that doesn't just go for which tools, but also how they are being used. Um it should also uh very clearly lay out how to report suspected misuse, um, who they should go to, what the chain of authority is. You should ensure obviously we've heard a lot about this, just that sensitive staff and student data is protected, including as already required by law.

SPEAKER_01 2:43

Yeah.

SPEAKER_00 2:44

And then finally, I would definitely recommend that district AI policies also incorporate AI literacy training as much as possible for staff and for students. So hopefully they're kind of building some instinct for how to use AI appropriately.

SPEAKER_01 2:59

So are most districts uh are they starting from scratch or are you hearing that they're just tweaking what they may already have?

SPEAKER_00 3:07

That's a good question. I would say it's hard to tell for a majority of districts what they're doing, but I definitely don't recommend just starting from scratch unless you know quite a bit about AI, then you know, go forth and conquer. But I think it's a better idea to start with Odoo's model or OSBA's model at the board policy level, just to kind of make sure that your bases are covered, and then customizing your local procedures to flesh out your framework. So I think the the nitty-gritty details can really live in your local procedures rather than board policy, so they're more flexible so that you can revise them more quickly since AI changes so fast.

SPEAKER_01 3:46

Uh let's talk about FERPA. Uh it's the student privacy law. Uh, how does uh the use of AI cause student privacy concerns for a district?

SPEAKER_00 3:55

Yeah, so this is crucial. Hopefully, districts have been hearing about this. Um if they haven't, obviously FER FERPA requires districts to keep student records and also personally identifiable information private. Um so think you know, name, birthday, identification numbers, things like that. Um but we know that AI companies obviously collect a lot of data, um, including anything you type into their AI tools frequently. So whether they use that to train their models or improve their products or just to kind of sell off to third parties. Bottom line is that districts should definitely know where that data is being stored, how it's being used, um, and where it's going uh in order to make sure that it remains verbate compliant.

SPEAKER_01 4:38

Aaron Powell Do you do you think schools realize how big of a deal uh that piece is yet?

SPEAKER_00 4:44

Aaron Powell You know, I think that if they're integrating AI literacy into sort of staff training, for example, they probably have at least heard of this kind of complication, but I can really only hope that they're taking it seriously.

SPEAKER_01 4:58

One area that feels like it's changing quickly is chatbots. Seeing a lot about this in the news, especially some of these uh companion or mental uh health type tools. Oh, yeah. Uh what concerns you most from a legal or student safety standpoint?

SPEAKER_00 5:16

Aaron Powell I have a lot of personal concerns, but since we're trying to stick to legal and student safety, I would say that first thing I'd mention is just that there have already been a lot of lawsuits against AI developers because of some of these chatbots. So often that's been in the context of um children, especially who end up hurting themselves after trusting these chatbots or treating them like a friend or like a romantic partner, like you said, even like a therapist. So I'm not aware currently of school districts involved in some suit yet, but I definitely see some risks. So I would just encourage districts to really think through those potential risks or even exposure to legal liability before introducing them.

SPEAKER_01 5:56

Um is this something districts should be limiting or just managing more carefully? How do you see that?

SPEAKER_00 6:04

I would honestly say that, you know, since some students may already kind of wrongfully trust these chatbots or engage with them in ways that are inappropriate, um student use of AI chatbots specifically should be approached especially cautiously. Um so there are some AI ed tools uh that may use chatbots that are still generally recommended to be safe for staff and student use. Um so some of those might be BRISC, uh school AI, and then also magic school. Um either way, I would just recommend AI tools that are designed for K through 12 over other um others for school use. So either way, um districts should review all of their AI tools, including chatbots, to see whether they were developed or modified with uh K through 12 in mind, whether they're FERPA and COPA compliant, and whether the AI provider can actually demonstrate that they're safe.

SPEAKER_01 6:59

So uh there's the misuse side. Uh deep fakes, uh AI-generated content. If a student uses AI in the wrong way, are existing discipline policies enough, or do districts need to think differently about all this?

SPEAKER_00 7:16

Yeah, I mean, this is a wild development. I mean, even if you if you remember some of the early AI from like 2023, it's just a world of difference. So that's another great question. Um OSBA is still working, obviously, on addressing this issue in policy, although it's kind of generally addressed in our model and in Odu's model as a form of harassment and bullying using AI. But it's a sensitive topic that district should definitely be aware of now, um especially deep fakes that are being developed in sexual contexts, which are now absolutely a legitimate trend in schools and across the country. So there was one case out of Pennsylvania where two male students at a private school were found criminally liable after creating hundreds of these deep fakes of over 60 girls. And then there are cases cropping up in Ohio, both inside and outside of schools too. So it's definitely not just theoretical. It's something that uh kind of we're being we're being hit with now. I would just say that the takeaway is that districts should be ready to address this issue before it ever becomes an issue in their community.

SPEAKER_01 8:23

So if you had to boil it down, what's the biggest legal risk for schools when it comes to AI right now?

SPEAKER_00 8:30

Um, you know, there's a there's a ton of places we could go. You could obviously talk about data privacy or over reliance or you know hidden uses of AI, but I would say that the biggest issue is that districts just don't know enough about AI. That's usually um I would say the issue that I run across. So AI literacy should be an absolute top priority for districts, um, just because you can't really protect against a legal risk that you doesn't know that you don't know exists.

SPEAKER_01 8:58

So, from a board member or superintendent perspective, uh what's the one thing they should do this week to be ready for that July 1st deadline if they're not already ready?

SPEAKER_00 9:12

Well, you know, you said one thing, but I'm gonna give you a couple anyways. First of all, um one thing I should definitely mention is OSBA's first AI workshop is now on demand. So I would love to plug that. Yeah, so all of those sessions from the April 11th uh workshop, which were awesome, are now available for purchase on our website. You can go to the training and events page and look for that on-demand learning option to see those. Um also, speaking of AI literacy, I I know we point to this resource a lot, but like Ohio's AI Toolkit, if you just want to start a place to start, um, whether you're a teacher or a parent or you want to actually draft policies and procedures on AI, it is a great primer. Um but as far as kind of practical next steps, uh just the things that I mentioned at the beginning, adopt an AI policy, ASAP, um figure out what tools are already being used in your districts, kind of create an inventory and decide how to choose between them. But then above all, maybe even as important as adopting that policy before the July 1st deadline, um, is just to pull together a technology committee that can actually dedicate time to track AI developments because it's only going to evolve faster from here on out, and you won't be able to track them on your own.

SPEAKER_01 10:26

And that's for sure. AI is not going anywhere. Uh absolutely. It's it's here to stay. Yeah, that's right. Got me. Thanks so much for your insights today. And uh, we will see you on the next Legal Edge sidebar.