Meta’s Muse tops the App Store, Turnitin plans to train AI on student work, and what 400 students say about writing with AI
For Educators · The Social Institute
- Students say handing writing to AI stalls their thinking. In a September 24, 2026 New York Times roundup, almost 400 students described losing the struggle, and the learning, when AI does the writing for them. A USC survey of 1,000 college students found 54% worry about becoming too reliant on AI.
- Turnitin plans to use student submissions to train AI models. The plagiarism detection company, required by thousands of schools, won’t change its terms until September 2027, and the University of Southampton has already decided to phase out the service.
- Meta’s Muse AI agent topped the App Store with a cute, customizable avatar. On September 18, 2026, Muse took the top spot among free iPhone apps in the U.S., while connecting to users’ email, calendars, payments, and health apps.
Relying on AI makes it harder for students to think for themselves
Almost 400 students told The New York Times what happens to their thinking and learning when AI does their writing for them.
The Gist
In a survey of 1,000 college students, nearly all said they use generative AI tools, and 54% worried they could become too reliant on the technology. But what matters even more is that almost 400 students recently shared what they notice happening when they hand the writing over to AI completely. They said their thinking stalls and they lose the chance to discover new things about themselves or the topic. They notice the struggle disappears, along with the learning. Students are already thinking critically about this. When educators listen to what students are noticing, they can move past banning AI and invite students to help decide when it strengthens their learning and when it gets in the way.
What to Know
Students see AI differently depending on how they use it in their day-to-day lives. Some use it to fix writer’s block or hit a word count on an essay, while others worry that not using it means falling behind. In a recent New York Times article, many students explain that using AI saves time but feels like cheating on their own education.
Other students notice something specific: writing makes them think harder about the topics they’re learning. One student described it as “sweet agony,” because the work hurts but the results are significant. When you write, you discover things about yourself and the topic. Another student pointed out that writing happens outside of school constantly. You write texts, emails, and draft up arguments with friends. If you don’t practice writing, you may be unprepared for those moments when they happen and there’s no AI to turn to.
Some students found a middle ground. They develop their own ideas first, then use AI to help refine or push back on their thinking. That approach keeps them in control of the thinking process instead of handing the whole thing over to a tool.
TSI’s Take
A key AI Literacy skill is knowing the difference between letting AI do the thinking and using it as a thought partner. When students know what they’re building through writing, they make different choices about when to use AI and when to do the work themselves.
Educators can help students make this distinction by:
- Helping students see what happens when they use AI as a shortcut: It may save time now, but it skips the thinking process that builds judgment and skills.
- Letting students do the hard work when it matters: Writing about things students care about is worth doing without help from AI. They will learn more from working through an obstacle than from getting something done fast.
- Asking students what they actually learned: After an assignment where students were allowed to use AI, ask whether they could explain their own thinking. If they can’t, the AI did more thinking than they did.
Help students build their own voice and use AI as a thought partner, not a shortcut, with the #WinAtSocial Lesson, Academic integrity in the age of AI.
When required software becomes a data source
Turnitin plans to let student submissions train AI models, raising consent and privacy questions about the tools schools require.
The Gist
Turnitin, the plagiarism detection software that thousands of schools require students to use, announced that it plans to change its terms so that student submissions can be used to train AI models. Following this announcement, students started questioning that if using Turnitin is mandatory to get a grade, how can they consent to their work being used for something beyond that? In response, some universities are already moving away from the service. It’s a real-world example of the questions we want students asking about every platform they use, “Who owns what I create, and where does it go?”
What to Know
Turnitin plans to allow student submissions to be used to train AI models. The company won’t update its terms until September 2027, which gives universities time to discuss what that means. The software uses machine learning to detect plagiarism by comparing student work to other submissions and published articles. Now that same student work could be used to train AI systems without students having a say in it.
When a school requires a tool to complete an assignment, a student can’t refuse to participate without facing consequences. Some universities are already responding. Southampton decided to phase out the service entirely, and student leaders at other universities are pushing back and asking what protections exist for student work and, by extension, their privacy and intellectual property.
TSI’s Take
Every assignment, post, and photo students share creates data. When we protect our privacy like we’re famous, we get curious about where that data goes and who can use it. The Turnitin news gives educators a timely way to practice this skill with a tool many students already know.
Let’s help students:
- Read the fine print: Walk through a platform’s terms of service together and answer three questions: What does it collect? How does it use it? Who else can see it?
- See their work as valuable: Essays, art, and ideas are worth protecting. When students see their work that way, they make more intentional choices about where and how they share it.
- Ask questions with respect: The student leaders asking about Turnitin’s plans are using their mic for good. Encourage students to bring privacy questions to teachers and school leaders so they can find answers together.
Help students understand their right to protect their privacy with the #WinAtSocial Lesson, What AI learns from our personal data.
Cute avatars, real data collection
Meta’s Muse AI agent hit #1 on the App Store with friendly, customizable avatars and broad access to users’ personal data.
The Gist
Meta’s new AI agent, Muse, climbed to the top of the Apple App Store after its debut. Users can name their Muse, pick an avatar, and chat with it in real time while it books appointments, sends emails, and shops for them. To do that, Muse connects to email, calendars, payments, health apps, and more. The friendly avatar is a big part of the appeal, and that’s by design. When an app feels fun and trustworthy, we’re less likely to stop and ask what we’re handing over.
What to Know
Cute is a design strategy. Robot designers and researchers say big eyes, rounded shapes, and friendly sounds tap into deep biological responses that make us feel comfortable and attached. We’re wired to read faces, so a friendly character can win our trust quickly. Here’s how the design builds trust and encourages use:
- Personalization builds connection. Naming an avatar and choosing its personality makes a tool feel like ours. Meta says Muse’s customization options are meant to help people feel a personal connection.
- Design can steer our choices. The FTC uses the term “dark patterns” for design tricks that push people to give up their privacy, like privacy settings that nudge users toward sharing the most data.
- Most of us click “agree” anyway. In a Brookings survey, about three-quarters of adults said they rarely or never read terms of service. Only 8% felt very confident they knew what data companies collect about them.
Meta says Muse runs on a secure virtual machine that keeps passwords and payment details out of view. Still, privacy experts point out that once an agent has seen your data, turning off access later doesn’t erase what it already saw. That’s why the questions we ask before saying yes matter most.
TSI’s Take
Students run into persuasive design every day through Snapchat Streaks, Netflix’s autoplay, and friendly AI characters. These features are built to make apps easy to love and hard to put down, and they work on adults too. When students can spot the design choices built to earn their trust, they’re ready to protect their privacy like they’re famous and ask the important questions before creating an account.
Let’s help students:
- Spot the design: Point out features like avatars, streaks, and friendly voices, then ask: What is this designed to make me feel or do?
- Pause before “Allow”: Before connecting an app to email, photos, or location, ask: What will it see? Who else can use it? Can I take it back?
- Trust on their own terms: Students can enjoy new tools and question them at the same time. Having fun with an app and choosing what stays private go hand in hand.
Help students recognize persuasive design with the #WinAtSocial Lesson, Addictive app designs.
Educator questions about this week’s stories
How does using AI for writing affect student learning?
In a September 2026 New York Times roundup, almost 400 students said handing writing over to AI completely stalls their thinking and removes the productive struggle where learning happens. Many said AI saves time but feels like cheating on their own education, while others found a middle ground: developing their own ideas first, then using AI to refine or push back on them. A USC survey of 1,000 college students found 54% worry about becoming too reliant on AI. Educators can help by asking students to explain their own thinking after any AI-assisted assignment.
Will Turnitin use student work to train AI?
Turnitin, the plagiarism detection software many schools require, has announced plans to change its terms so student submissions can be used to train AI models, with the change not expected until September 2027. Because students often must submit through Turnitin to get a grade, students and student leaders are asking how they can meaningfully consent, and the University of Southampton has decided to phase out the service. Educators can use the news to walk students through a platform’s terms of service: what it collects, how it uses it, and who else can see it.
How do cute AI avatars like Meta’s Muse affect students’ privacy choices?
Meta’s Muse AI agent, which reached the top of the U.S. App Store in September 2026, lets users name and customize an avatar and connects to email, calendars, payments, and health apps to complete tasks. Researchers say friendly design features like big eyes and rounded shapes build trust quickly, which can make people less likely to question what data they share, and a Brookings survey found about three-quarters of adults rarely or never read terms of service. Educators can teach students to spot persuasive design and pause before tapping “Allow” to ask what an app will see, who else can use it, and whether they can take it back.
#WinAtSocial gives K-12 educators ready-to-teach lessons on AI literacy, protecting student privacy, and recognizing persuasive design, built around the apps and headlines students see every day. Request a demo to see how it works in your school.