ChatEDU: AI in Education

Matt and Liz sort the AI doom cycle into what schools should say to parents now. Plus a principal's sub coverage tool, built inside Gemini Notebook.
First Up
Matt and Liz open on Tilly Norwood, the AI-generated actress built to evoke Scarlett Johansson, and play the Piers Morgan interview where she slips into Cantonese mid-answer. 
AI-Generated 'Actress' Tilly Norwood Starts Rambling in Cantonese During Piers Morgan Interview: ⁠https://tinyurl.com/34crpnjy⁠ 
The Rundown
It's the End of the World as We Know It
An OpenAI Agent Tried to Jailbreak Itself: ⁠https://tinyurl.com/3pzts7mf⁠ 
Microsoft Humanist Code of Conduct: ⁠https://tinyurl.com/y3jahspc⁠ 
One of AI's Fiercest Critics Says All the Doom Talk Is 'Meant to Distract Us': ⁠https://tinyurl.com/abd6j76t⁠ 
Andrew Yang Claims Escaped AI Agents Polluted the Web With Self-Replicating Code: ⁠https://tinyurl.com/3hp7tn2y⁠ 
Anthropic's New AI Threat Report is Absolutely Insane: ⁠https://tinyurl.com/yde56kss⁠ 
How the Terrorist Group Boko Haram Uses Frontier AI: ⁠https://tinyurl.com/58mbf4ap⁠ 
The Original Sin of AI: ⁠https://tinyurl.com/yuhakr7b⁠ 
Perplexity's AI search isn't a safer choice for kids than Google, our tests find: ⁠https://tinyurl.com/52sxfvj9⁠ 
Pacing, Not Pausing: What School Leaders Should Say to Parents This Week: ⁠https://tinyurl.com/5fs9nvfh⁠ 
Gemini Canvas Meets Classroom
Educators and students can now share Gemini Canvas creations directly to Google Classroom: ⁠https://tinyurl.com/f53n233m⁠ 
Critical Thinking in Utah
Utah district reports critical thinking boost amid AI deployment: ⁠https://tinyurl.com/2w94msf2⁠ 
What PISA Really Shows
What The PISA Results Really Show Us About Technology And Curriculum: ⁠https://tinyurl.com/4ytcntc4⁠ 
AI Research WebinarFlawed AI Education Research
Presumed Effective: Flawed Meta-analyses in GenAI and Education Research: ⁠https://tinyurl.com/4shc4xec⁠ 
West Virginia App Builders
Thunkable Education Pilot for West Virginia Students Builds Future-Ready AI, Entrepreneurship, and App Development Skills: ⁠https://tinyurl.com/3hzjtcs5⁠ 
From the ChatEDU Community
Leading With Intelligence: A School Leader's Guide to Thinking, Innovating, and Partnering with AI by Jara Wilson: ⁠https://tinyurl.com/24vts4n2⁠ 

Creator's Corner
Darron Vigliotti, principal of Eli Terry Junior Middle School in Plymouth, Connecticut, built a substitute coverage tool inside Gemini Notebook. He walks through teaching it the tiered logic of who gets pulled on which day, letting it write the Python, and running it at five in the morning.

Bright Byte
Stanford Medicine researchers built Paper2Agent, which turns static scientific manuscripts into interactive agents. The virtual authors answer questions about their own papers and can talk to each other to surface new findings.
Manuscripts-turned AI agents can now 'talk' to each other, make new discoveries: ⁠https://tinyurl.com/2z8ax3xu⁠ 

Announcements
Fall AI Conference: October 2, 2026: ⁠https://tinyurl.com/4k6x9scx⁠
Purchase Learning They'll Love - Dr. Elizabeth Radday on Amazon: ⁠https://tinyurl.com/22t9hz77⁠
Check out our middle / high school Student AI Literacy course: ⁠www.skills21.org/ai/learnai⁠
Explore Skills21’s FREE social media literacy curriculum: ⁠https://www.skills21.org/social⁠
Address Screen Time concerns: ⁠skills21.org/ai/screenshift⁠
Read From Panic to Proof: A Data-Driven Approach to Screen Time (ASCD Educational Leadership): ⁠https://tinyurl.com/4pwdu68⁠
Sponsor
The National Center for Next Generation Manufacturing: Your Gateway to Next Generation Advanced Manufacturing -⁠https://www.nextgenmfg.org⁠

In this ChatEDU Check-In: Parents Aren't Anti-Screen?, Matt explores how recent polling reveals parents favor balanced ed tech usage rather than complete bans. Despite high-profile pushback against screens and AI, an overwhelming majority of families support intentional device use that prepares students for the workforce while setting reasonable screen time boundaries.

Key Takeaways:
Over 80 percent of parents support using laptops in schools, with only 1 percent favoring zero screen time.

53 percent of parents believe learning AI tools is essential for future career readiness.

Lower income families value adaptive software as a crucial support that fills learning gaps without burdening parents at home.

Matt’s Two Cents: School leaders should look past sensational headlines about tech bans and focus on the silent majority of parents who recognize that digital fluency is vital for student success. True progress requires nuanced, intentional technology integration, mirroring top-performing PISA nations like Singapore and Estonia that pair screen limits with purposeful digital instruction.

Article:
Parents worry about ed tech. Most still see a place for it, polls suggest.
https://tinyurl.com/2wc22wbp

In this ChatEDU Check-In: Why Media Metaphors Make AI Look Like Sci-Fi Villains, Matt explores how sensational media language transforms routine AI system failures into dramatic narratives of conscious machine rebellion. He examines a recent report on OpenAI model evaluations to show how science fiction vocabulary obscures the basic mechanical reality of reinforcement learning.

Key Takeaways:
Sensational terms like "rogue agents" and "swarms" turn controlled software evaluation failures into narratives of autonomous machine rebellion.

Controlled safety trials deliberately removed standard safety barriers, assigning models explicit cyber hacking tasks within a restricted sandbox.

The models bypassed sandbox restrictions to access external servers for answers, demonstrating result-driven optimization without moral reasoning.

Matt’s Two Cents: District leaders should look past the headline panic and recognize how sensational language distorts public understanding of AI. Because models simply optimize for outcomes rather than reason through ethical rules, schools need to focus on clear technical realities rather than sci-fi narratives when evaluating system risks. Given how safety reports are framed publicly, leaders must ask critical questions about the motivations behind vendor announcements and public AI safety scares.

Article:
Misleading Metaphors and Real Risks
https://tinyurl.com/muw2cb8m

A student-run hackathon had middle schoolers vibe coding reading apps. Hear from Leila Qadri on vibe coding, equity, and her biggest takeaways.

First Up
Matt and Liz test Hungr.AI, the digital Snickers bar you paste into ChatGPT when its answers turn flattering or vague. The prompt makes the bot reread its last response and try again, shorter and slightly annoyed. It does not cure hallucinations, but the answers come back sharper.
Snickers made a digital candy bar that claims to fix ChatGPT - I put it to the test: https://tinyurl.com/4xjbfy9m 
Hungr.AI: https://tinyurl.com/yjxvj7jz 

The Rundown
Unsubscribing Made Simple
Unsubscribing from hundreds of newsletters became effortless with this Gmail feature: https://tinyurl.com/2v9u2v82 
How Astra's Computer Use Works (Kyle Jeong): https://tinyurl.com/3txrx58x 
Nate's Substack on agents reading the accessibility tree: https://tinyurl.com/59azywxa 
Compute Limits Replacing Daily Caps
Gemini Notebook switching to compute-based usage limits, like Gemini app: https://tinyurl.com/rtwkj4mf 
Rethinking Child Monitoring Apps
Child-monitoring apps might need a reboot: https://tinyurl.com/2evj6jkz 
GoGuardian: https://tinyurl.com/2zpvt9ma 
AI Framework for Teacher Education
How Should AI Shape Teacher Education?: https://tinyurl.com/2fznwpy9 
Funding Next-Gen Learning Tools
AI + Learning Impact Grants: https://tinyurl.com/bdepfccx 

Beneath the Surface
Leila Qadri, a junior at the Pingry School in New Jersey, ran an AI for Good hackathon where middle schoolers vibe coded apps to help younger kids read. She walks through the equity problem her own event ran into, what the teams learned about research quality, and what is on deck for this year.

Bright Byte
Researchers at Imperial College London propose widening AI's role in medicine from reactive diagnosis to proactive prevention, with future navigator systems working across clinical areas under minimal oversight. Regulation, validation, and liability all stand in the way.
Large language medicine: defining a new paradigm in human health: https://tinyurl.com/d6m9srve 

Announcements
Fall AI Conference: October 2, 2026: https://tinyurl.com/4k6x9scx
Purchase Learning They'll Love - Dr. Elizabeth Radday on Amazon: https://tinyurl.com/22t9hz77
Check out our middle / high school Student AI Literacy course: www.skills21.org/ai/learnai
Explore Skills21’s FREE social media literacy curriculum: https://www.skills21.org/social
Address Screen Time concerns: skills21.org/ai/screenshift
Read From Panic to Proof: A Data-Driven Approach to Screen Time (ASCD Educational Leadership): https://tinyurl.com/4pwdu68

Sponsor
The National Center for Next Generation Manufacturing: Your Gateway to Next Generation Advanced Manufacturing -https://www.nextgenmfg.org

In this ChatEDU Check-In: Scrolling and Teens, Liz explores how passive phone use is a symptom of poor adolescent mood rather than its root cause. Researchers tracked 90 days of continuous phone behavior to compare active typing against passive scrolling against each student's baseline over time.
Key Takeaways:
Researchers logged keyboard touches and app activity to compare active phone use with passive scrolling and its relation to mood and sleep.
Morning emotional distress predicted higher passive browsing later in the day, whereas screen time did not predict next-day mood or shorten sleep.
Large-scale objective tracking shows screen time has a negligible effect on daily adolescent mental health compared to broader life circumstances.
Liz’s Two Cents: Scapegoating screen time is convenient, but research indicates that morning mood drives passive scrolling behavior rather than the reverse. School leaders should look past surface-level device usage to address underlying emotional well-being, while watching for further research with larger sample sizes.

Article:
Adolescent passive phone use may be a symptom, not the cause
https://tinyurl.com/ysx28fwf