So many things have happened in the past week that impact schools: lots of Gemini Notebook updates, ChatGPT for Teens’ harsh review from Common Sense Media, suspected freezer hacks at US military bases and confirmed ones in France, a university’s chief academic officer resigning over undisclosed AI use in published materials, and a new heart study that our coaches need to hear about.
I also recorded some bonus material for “geeks” or more advanced people that you can listen to for the last three minutes if you like. Oh, and much more. I ended up with 17 news stories plus 7 new features from Gemini Notebook and some notes for administrators about the repercussions of these stories.
AI isn’t the first change to hit the world, but it seems to be the fastest. AI is impacting everything but it is more than AI being discussed this week. Every source is linked in the show notes.
Also, at the end of this, I’m breaking down all of the ways I use (and don’t use) AI as I created content for this show. I can’t really talk about disclosure of AI use without disclosing that sort of thing as the capabilities of AI are getting better by the day and I will admit, I really don’t like creating graphics but now that I can train for the graphics to look like those I create, I feel a little better about saving that 30 minutes a day creating graphics. As always, I want to hear what you think. (Oh, and I do share all my sources of information for this show which kind of shows the human in the middle of the creation process since I even pulled some news articles I learned about on my 5 am walk this morning.)
In short: AI in schools news moved fast this week. Teachers can now see student chat transcripts in Gemini Notebook, OpenAI retires custom GPTs on December 11, and a Dartmouth provost resigned over undisclosed AI use. The common thread: students and teachers need clear rules about privacy, disclosure and what AI should and shouldn’t do.
Seventeen stories, numbered to match the chapters and the research table below. Click a question to jump to that spot in the transcript.
Here’s the vocabulary definitions. Research is included in the research box below.
Here’s where the facts in this episode come from, with the time stamp so you can hear each one. Checked November 3, 2026. Story numbers match the chapters.
This transcript was generated using AI and has been reviewed by humans for accuracy. Minor errors or artifacts may remain but I worked my best to find any issues with the transcript as I reviewed the show. – Vicki
Vicki Davis (00:00): Welcome back, educator. Today is Tuesday, November 3rd, 2026, and this show is a quick news update on the technology and news that impacts schools. All the links are in the show notes at coolcatteacher.com forward slash E1022. Today. Freezer hacks, Gemini Notebook updates, why we can’t trust AI meta analysis yet, custom GPTs are going away, AI agents, energy drinks and your heart, AI disclosure, and two senators write a letter to Google from both sides of the aisle.
Vicki Davis (00:37): Stay tuned at the end of the show for a bonus for those of us who are a little more geeky about how some of your vibe coding may break if you’re using Claude and how to fix it, and how some Discord server chats are going public at the end of 2026.
1. Why are schools being told to update freezer and HVAC firmware?
Vicki Davis (00:54): This one is for your kitchen crew. Update your freezer firmware. Military Times reports that the Pentagon is treating freezer malfunctions that spoiled food in at least six military-based commissaries in August as a suspected hack.
Vicki Davis (01:11): In October, a cyber attack overrode freezer temperature controls in more than 350 establishments in France. And NPR reports refrigeration hacks in Israel and other parts of Europe, too. Danfoss, whose controllers ran many of those French units, says that many were running outdated firmware.
Vicki Davis (01:31): Okay, firmware matters and it’s everywhere. So update that and take those devices off the open internet if you can. But this is not just freezers. Do it for everything that a technician might dial into. In schools, this could be security cameras, door access, HVAC, and lighting controls, vending machines, and projectors. Again, follow the resources. This is just supposed to be an overview of news and not recommendations for everything you need to do.
2. What’s new in Gemini Notebook for teachers?
Vicki Davis (02:01): So here’s some good news. Google Notebook LM was renamed Gemini Notebook on July 16th, 2026. Big news last week. The phone apps can now record up to one hour of live audio and turn it into a source. Now, one note, it keeps the transcript, not the audio.
Vicki Davis (02:21): Record the class. You can turn into notes and study guides right away. But remember, it captures student voices. I would recommend that you turn it off for class discussions and check your district’s recording policy.
Vicki Davis (02:33): If you’re enabling this for your teachers, you probably want to have that conversation now. Since October 19th, 2026, teachers can see summaries of how students use notebooks assigned in classroom, including our individual student chat transcripts. Now, students get notified about this, but tell them too.
Vicki Davis (02:52): Kids think that what they tell AI is private and it’s not. Now, this only works for notebooks you assign inside Google Classroom. So make the notebooks for your classes. I highly recommend this. Now, coming soon to our notebooks, connectors.
Vicki Davis (03:10): When it arrives, I highly recommend that you link Google Drive. That would be a no-brainer. And hopefully, at some point, they’ll be able to let us create those quizzes and turn them into Google Forms easily. In the Gemini chat, type @Gemini Notebook to create notebooks inside Gemini chat and add sources and ask questions.
Vicki Davis (03:33): Now, since October 29th, Gemini lets you edit the docs, sheets, and slides that it makes and download slides as a PowerPoint. Here’s the only catch. In Gemini Notebook, any slide revision will make a whole new deck and exports don’t sync back from Google Drive back into the notebook.
Vicki Davis (03:51): So you have to re-import those. October 28th, Immersive Explorations was released in the Gemini app. About 20 interactive visual lessons from Silk Road Trade Routes to the Physics of Lightning. Look for Try Immersive Explorations in the Student Hub.
Vicki Davis (04:08): And October 29th, hundreds of free textbooks in Gemini Notebook from OpenStax, the Open Textbook Library, and S. Chand. Click Add Sources, then Textbooks, and Answers will actually cite the page in the textbook.
3. Can we trust meta-analyses of AI in education?
Vicki Davis (04:22): Second, AI research is a mess. I attended a session with Justin Reich from MIT. EdWeek’s headline about the session, since they obviously had somebody in there too, the research on AI and education is a mess. Okay, that’s putting it mildly. So Justin made a comment that really stuck out to me that gives us perspective.
Vicki Davis (04:42): So he has two researchers he really likes to follow in terms of ed tech. Harvard physics researchers Greg Kestin and Kelly Miller. He says they can really get things out quickly and with excellence. Their first study came out in June 2025. Here’s what Justin said.
Vicki Davis (05:00): If it takes them till 2025, it’s going to take the rest of us till 2030. And there’s no reason for meta-analysis for years afterwards. Now, I’ve had him previously on the show. I will link to him in the show notes. He had the interview with Patrick O’Neill of Ivy Tech Community College, who did a forensic analysis of 14 meta-analyses of AI in education.
Vicki Davis (05:22): And Patrick found impossible statistics and likely hallucinated references. And ChatGPT studies dated before ChatGPT was released. According to O’Neill’s research, about 65.8% of conferences call for papers, explicitly asking for positive AI findings.
Vicki Davis (05:43): And about 20% favor them. Only about one in seven was neutral. So AI in education is nuanced. There is huge room in the middle for what Cal Newport, in his latest podcast, calls techno-selectionism.
Vicki Davis (06:00): That means that you’re not all a fan, you’re not totally against, but you know that you can select tech for good uses and avoid it for uses that you don’t think are good. Now, Justin’s tip at the end of this was to read a study as a hypothesis and test it in your own classroom.
Vicki Davis (06:19): And we have to watch for something called de-skilling when students lose skills because AI does the work.
4. How many students use AI during quizzes and tests?
Vicki Davis (06:27): Now, I had Dr. Brian Fitzpatrick in an interview this morning as I’m recording this, and he will be on an upcoming Cool Cat Teacher Talk, and 10 Minute Teacher episode. As we talked about some research from the Fordham Institute, that was extremely alarming.
Vicki Davis (06:46): A significant number of our students are actually using AI as they take quizzes and tests, right around 25%.
5. What happens to custom GPTs when OpenAI retires them?
Vicki Davis (06:53): Okay, more on that later. OpenAI is retiring custom GPTs on December 11th and replacing them with plugins. Harvard says its users have until February, so I’m not quite sure the difference there, but okay. Download your knowledge files now, and if you want to stay on ChatGPT, use Migrate to Plugin.
Vicki Davis (07:15): Test it with prompts you know, and regenerate your share link if you’re using it with students, because old links will break.
6. What are always-on AI agents like OpenAI Dots and Claude Duet?
Vicki Davis (07:23): Next, always-on AI agents are making the news. OpenAI launched Dots. These are agents that keep working after you close the chat. They are not in the free K-12 teacher plan yet. Claude Duet is for pro and max subscribers in the U.S. I’m testing it, and I will say it’s not ready for prime time yet.
Vicki Davis (07:42): But here’s the biggest issue I see coming. Student data. Duet has a dedicated classifier for sensitive information, but it sits outside your district’s data agreements. So be super careful if you’re testing this on personal accounts, not to have an issue with that.
Vicki Davis (08:00): So voice is going to be a way that we interact with AI, and it’s going to get tricky with that PII. I had a recent show with Rachelle Dené Poth on this topic. Take a listen to her and other experts about how to handle PII in your school.
7. Are teens using AI chatbots for emotional support?
Vicki Davis (08:17): NPR reports on a JED Foundation survey of more than 5,000 tweens and teens, and 58% use chatbots for schoolwork, and a similar number when they’re sad, stressed, lonely, or overwhelmed. This should be of concern to us using chatbots for companionship or when distressed was linked with more loneliness and less resilience.
Vicki Davis (08:36): And researchers cannot tell which comes first. UNC’s Anne Maheux says it’s likely both. If you or a student is struggling, call or text 988.
8. Are energy drinks dangerous for teen athletes?
Vicki Davis (08:49): And this research came out yesterday on November 2nd, 2026. The American Heart Association released preliminary research reviewing more than 28,000 adverse event reports from 2017 to 2025 for ages 13 and up.
Vicki Davis (09:05): Compared with other products, caffeinated energy drinks and stimulant supplements, were linked to cardiac arrest about 11 times more often and heart attack about four and a half times more often. About 45% of the heart-related reports involved adults 18 to 34.
Vicki Davis (09:25): Now, this is a link, not proof of cause. It hasn’t been peer-reviewed yet. It’s being presented this week at the American Heart Association Scientific Sessions. But talk to your athletes and older students about pre-workout supplements and energy drinks. And I will recall what my friend Lisa O’Connor did. She was a middle school science teacher and also worked with high school health classes.
Vicki Davis (09:48): They did an amazing multi-day lab measuring all the different things about energy drinks that really did change how students behaved with them when I saw her do that.
9. Why did Dartmouth’s provost resign over AI use?
Vicki Davis (09:59): Talk to your students about AI disclosure. Dartmouth’s provost, Santiago Schnell, resigned after a faculty-led committee found he used AI to meaningfully alter his writing without disclosing it. This was not with AI detectors. They reviewed his university Claude logs and his email.
Vicki Davis (10:17): He had used Claude to edit a research manuscript, troubleshoot code, and generate figures. But here’s the irony, he wrote a Washington Post op-ed about AI cheating. This is a model for what’s coming. It’s okay if you wrote it, prove it. And it’s okay if you use AI, disclose it.
10. Why did UNC-Chapel Hill stop using AI to score admissions essays?
Vicki Davis (10:37): In other AI news, UNC Chapel Hill will stop using an AI tool that rated application essays. So many applicants now polish their essays with AI that the scores rose until it couldn’t tell applicants apart. We see that AI often removes the best of student voices, help students tell their own story when they can back up with an interview.
Vicki Davis (10:57): There’s a difference between using AI to fix your grammar and spelling and give you feedback, and one that completely obliterates your voice.
11. Can an employee get a religious exemption from using AI?
Vicki Davis (11:05): David Durman, a research administrator at the University of Kentucky, and non-theistic humanist, received a religious accommodation from using AI at work. Okay, I find this interesting because this is my take. Is he not going to use Google search? Spellcheck?
Vicki Davis (11:23): How about Netflix, which has recommendations? To me saying you won’t use AI is like saying you won’t breathe air. I think he probably should be more specific and maybe he has been. I won’t use an LLM, but you can expect this to travel.
12. What did senators ask Google about AI in schools?
Vicki Davis (11:39): On October 21st, Senators Blumenthal and Hawley sent Google nine questions about AI in classrooms due back November 2nd. That was yesterday. Their concerns, AI on by default and kids depending on it for schoolwork and emotional support. In my own admin duties, when things got turned on, it took me hours to figure out how to turn them back off.
Vicki Davis (12:03): And if you have anything on inside Google, check Google Vault so you can see what students are saying to AI.
13. Is ChatGPT for Teens safe for kids?
Vicki Davis (12:13): Now, new numbers from Common Sense Media. 86% of kids aged 9 to 17 now use AI. Another one we should really be concerned about. Common Sense Media rated ChatGPT for teens an unacceptable risk in October. That’s its lowest rating, and that is a step below high risk for ChatGPT.
Vicki Davis (12:37): Parent alerts and crisis referrals often didn’t work in their test. Now, OpenAI is disputing the findings and wants a retest, but our kids are using AI, and from what we’re seeing, it’s not safe. We live in a pace of rapid change, and right now on Cool Cat Teacher Talk, I’m running a series called Stop Pretending.
Vicki Davis (13:02): The first one was Stop Pretending, Start Learning, and it was about learning. Second was to stop pretending we’re not addicted to the algorithm and how to get our life back. Now, after our outro, if you’re nerdy about AI or just kind of a geek or curious, I have some bonus content for you.
Vicki Davis (13:21): If you subscribe to the 10 Minute Teacher, subscribe to Cool Cat Teacher Talk, too, for longer interviews and many voices on one topic. That’s my weekly radio and TV show. I also post it to YouTube. All the links today are at coolcatteacher.com forward slash E1022.
Vicki Davis (13:40): Thank you for supporting the show, and here’s our bonus content.
14. Why won’t Claude Code recreate a game I name?
Vicki Davis (13:45): First, for teachers who are vibe coding, Anthropic added a rule to Claude Code. On Opus 5.5, it won’t recreate someone else’s specific game, even if you rename it. So people were going in and saying, create Mario Kart or create this or create that. The problem was that they were concerned about intellectual property.
Vicki Davis (14:07): So developers have reported some reverse engineering blocks, and Anthropic says it’s still adjusting. I was listening to this on Tech Brew Ride Home. Now, this could impact teachers who vibe code games for kids based on other games. So if you’re using Opus 5.5, you might need to describe the game a little differently.
Vicki Davis (14:29): If you have trouble, some say that changing the model from Opus 5.5 might help you.
15. What is wolfmasking scam training?
Vicki Davis (14:36): Next, wolf masking. I learned about this in a show from Brian Brushwood, host of Scam Nation. This is supervised role play where people practice being a scammer. But here is the way I’ve heard it described. The payload is something harmless like a cute dog picture.
Vicki Davis (14:52): As Brian puts it, the sheep need to learn how to play the predator in safe environments. I do hate to say we might have to get to this point.
16. How is AI changing school cybersecurity risks?
Vicki Davis (15:00): With all of the AI hacking going on, like the October 21st PHP shell problem, the language behind roughly 70% of websites where it can be measured. They rushed out emergency patches for two critical flaws. AI helped find them, and then the news leaked before the patch.
Vicki Davis (15:18): The Pragmatic Engineer, Gergely Orosz, warns that AI makes exploiting flaws easier for everyone, and it’s only a matter of time until students do it. Perhaps this is another reason to have fewer devices and go back to computer labs. I’ve always loved computer labs.
Vicki Davis (15:38): So patch your school servers now.
17. Will public Discord messages show up in Google search?
Vicki Davis (15:40): And finally, Discord is going public. Publicly listed Discord servers with at least a thousand members look like they’re going to be getting public webpages starting November 16th. Now they’re saying there’s options or there may not be options.
Vicki Davis (15:53): I don’t really understand it all. But starting in mid-December, public channel messages may show up in Google without usernames. DMs, private servers, older messages, and teens are supposed to be excluded. Now, this is on by default in the U.S., so to turn it off, go to user settings and then data and privacy and turn off channel messages.
Vicki Davis (16:13): I do expect this to impact schools because our kids are all over Discord as well as our gamers. So talk to your staff and older students and check your school and club servers. So thank you for listening to all of this content. I do plan to do this every so often. I will share them when I have news. See you later, educator.
Announcer (16:33): Thank you for tuning in to the 10 Minute Teacher Podcast. Join us here every weekday and subscribe to the Classroom Matters newsletter. See you later, educator.
I believe in being transparent about how I use AI, so here is exactly how this episode came together: what I did, what AI did, and where Claude and I checked the facts. I co-authored this with Claude as I had Claude examine the chat used to create this show and to fact check against what Claude and I did and to provide specific examples. I wrote much of this but allowed Claude to craft some explanations from the chat observations. I often do this to deconstruct how I used AI and what the AI tool did, no matter the tool I use.
I found these stories the way I always do. I listened to podcasts on my morning walk, read through my Feedly feed reader, and caught things on the news and on YouTube. When I had a topic I wanted to cover, I pointed Claude to it and asked it to research.
Sometimes that turned up something I didn’t know. For example, I already knew Gemini Notebook could record audio, but I didn’t know it had added textbooks. When Claude pointed that out, I asked for the links and read the information myself before I decided to use it. There were some new features of Gemini Notebooks that were not as important to me that I didn’t include. For example, I skipped Gemini Notebook’s new admin sharing settings for trusted domains because I wanted to focus on what teachers would notice.
I drafted much of the script myself. Then I worked with Claude to condense it into a quick-hit news format and to make a teleprompter version with pacing cues. I edited every version, cut items that didn’t belong, and added my own tips. My athlete tip and my Opus 5.5 tip were both mine.
With just about everything I use Claude Docs. That means it takes my content and puts it into an editable document so that I can highlight and chat with Claude, fact check, and improve the content. This creates a full trail on how I use AI or not.
Anything Claude writes in my documents goes in [brackets] so I can quickly see it. If I approve it, I tell Claude to remove the brackets. That way, if I leave and come back to a job, I know what I’ve read and approved. This is a little workflow nuance, but it works for me.
I clicked every link and read the articles. I took every screenshot you see on screen myself, before I recorded so that I wouldn’t waste time talking about something I would have to remove later. I like to look at links before recording as I’m writing, for obvious reasons.
Claude fact-checked the script against the original sources, and some things changed because of it:
The list above was an itemized list of things that were changed as a result of the fact check and things that were verified so you can see an example of what was done.
I recorded the episode in Riverside. Riverside’s AI made the transcript. John used two of Riverside’s AI audio tools: Magic Audio to clean up the sound, and filler-word removal.
John, my editor, listened and edited the episode twice. I listened through the whole show three times and made changes every time.
When I recorded, I misspoke Cal Newport’s term “technoselectionism.” Instead of re-recording, I fixed it myself with Videodub inside Riverside. I highlighted the word in the transcript and typed the correct one, and it regenerated and changed “Technoselectionist” to “technoselectionism”. It only changes one word at a time, and it only works because I gave it permission to use my voice. I meant to say the correct word and misspoke and had written it wrong into the script. That was on me but this saved me time as the word was essential to the thesis of that section.
Claude made the on-screen graphics from my screenshots: source cards numbered to match the links in the show notes, a checklist for the Gemini Notebook features, and a definition card for PII (personally identifiable information). After creation, I opened all of the graphics and read them inside Apple’s Preview program. After verification, I uploaded them to Riverside. (I don’t let Claude upload anything to my public platforms.)
Then Claude placed the graphics at the proper time stamp the edit according to the transcript and moved my video to the side so you can see the source while I talk. I reviewed the graphics and asked for changes for three of them and the creation of two new graphics (a definition of PII and a card pointing to Rachelle Dené Poth’s earlier show). Additionally, I had one graphic recropped because it looked like an ad, and I replaced a topic card with the checklist so it matched what I was saying.
After all of this is done and I’m happy with it, Claude does a retrospective. First, it puts the small housekeeping fixes in place, like filing rules and notes. Then it talks about where there was friction (including its own mistakes), any errors in the skills I’ve written, and any recommendations for efficiency. If I see that a skill really does need an update, Claude writes the update to the skill in a Claude Doc, I edit it, and then Claude gives me the updated skill to save right then so I don’t have that problem again. This way, everything gets better continually.
AI didn’t choose the stories, decide what matters for teachers, or have the last word on anything. Every story, opinion, and recommendation in this episode is mine. AI helped me work faster and helped me double-check the facts.
Thank you for supporting the show! If you like my show, I’d love for you to leave a review. I guess you can leave a review if you don’t, that is your prerogative, but if you find it helpful, I’d really appreciate it!
