New AI Safety Framework Aims to Protect Kids in Schools
Researchers propose a smarter way to keep students safe from harmful AI content, but parents and teachers still need to stay alert.
More schools are turning to artificial intelligence (AI) to help students learn, but a new study warns that the technology still poses risks. On November 17, 2026, researchers published a proposal for a new safety framework—called EGS—designed to make AI tools safer for kids in classrooms. While the system aims to guide students away from dangerous or inappropriate content, experts say it’s not a cure-all. For families and teachers, the message is clear: don’t trust AI to keep your kids safe just yet.
What Happened
The research, posted on the preprint site arXiv, introduces EGS (Educational Guidance System), a set of rules and tools meant to make large language models (LLMs) like ChatGPT safer for students. Unlike older safety systems that simply block risky questions, EGS tries to understand why a student is asking and offers helpful, age-appropriate responses instead of just saying “no.”
AI chatbots are already being used in schools across the U.S., with companies like OpenAI, Google, and Microsoft pushing their tools into classrooms. The hope is that AI can personalize lessons and help students who might otherwise fall behind. But recent incidents—like chatbots giving out biased or inappropriate answers—have raised alarms. The EGS framework is one of the first attempts to address these concerns with a system built specifically for education.
Still, the researchers admit their work is in early stages. There’s no guarantee that EGS will catch every problem, and it hasn’t been widely tested in real classrooms yet. For now, parents and teachers are being urged to stay involved and not assume AI is automatically safe for kids.
How This Affects Everyday People
If you’re a parent, teacher, or student, the rise of AI in schools is already changing daily life. AI chatbots can help with homework, explain tough concepts, or even tutor kids one-on-one. But there’s a catch: these systems sometimes give out wrong, biased, or even harmful information. Imagine a student asking an AI about a sensitive topic—like mental health or relationships—and getting an answer that’s misleading or inappropriate for their age.
EGS is meant to help by guiding students toward safer, more helpful responses. For example, instead of just blocking a question about self-harm, the system might offer resources or suggest talking to a trusted adult. But because the technology is new, there’s still a risk that something slips through. That means parents and teachers can’t just hand over the reins to AI. They need to keep an eye on what kids are asking and what answers they’re getting.
For students, this means learning to be skeptical about what AI tells them. Just because a chatbot sounds confident doesn’t mean it’s right—or safe. And for families worried about screen time, AI tools can make it even harder to set limits, since they’re often built right into homework assignments or classroom activities. For more on juggling AI and screen time, see our guide on AI and screen time for kids.
The Bigger Picture
This new framework comes as lawmakers and regulators are starting to pay closer attention to AI in schools. In the U.S., the Children’s Online Privacy Protection Act (COPPA) already limits how companies can collect data from kids under 13, but it doesn’t cover the kinds of content AI chatbots might generate. Meanwhile, the European Union’s AI Act, passed in 2024, puts stricter rules on “high-risk” AI systems used with children, but enforcement is still catching up.
Major tech companies are investing billions in educational AI. Microsoft, for example, has committed over $1.5 billion to AI-powered classroom tools since 2023. But as these systems spread, so do concerns. In 2025, a widely reported incident saw a popular AI tutor give a 12-year-old student advice that was flagged as inappropriate by teachers. Stories like this are fueling calls for stronger safeguards and more transparency about how AI decides what to say to kids.
For families, the stakes are high. The push for AI in education isn’t going away—schools are under pressure to keep up with technology, and companies are eager to sell their products. But as this new research shows, the tools to keep kids safe are still playing catch-up. For more on how to keep your family safe, check out our advice on keeping kids safe from AI.
What You Can Do
- Stay involved: Ask your child what they’re using AI for at school, and look at the answers they’re getting together. Don’t assume the technology is always right.
- Talk to teachers: Find out what AI tools are being used in your child’s classroom and how safety is being handled. If you’re concerned, ask how teachers monitor AI use and what training they’ve had.
- Set boundaries: Make clear rules about when and how your child can use AI tools for homework or study. Consider using parental controls or monitoring software if you’re worried.
- Teach critical thinking: Remind your child that not everything an AI says is true or safe. Encourage them to double-check important information with a trusted adult or reliable source.
- Stay informed: Keep up with the latest research and safety tips. For more on what skills kids will need in an AI-powered world, see our guide on what skills kids need for the future.
The Bottom Line
AI is changing how kids learn, but the technology is still far from foolproof. The new EGS safety framework is a step in the right direction, but it’s not a silver bullet. Parents, teachers, and students need to stay alert, ask questions, and keep the conversation going about what safe, responsible AI use really looks like in the classroom. As the technology evolves, so will the risks—and the need for real human oversight.
AI Regulation Groups Outspend Tech PACs, Shifting the Rules
AI regulation groups spent more than industry PACs before the 2026 midterms, signaling new laws that could impact jobs, privacy, and daily life for regular people.
Read analysis CONTENT QUALITYAI Assistants Quietly Omit Key Warnings, Shaping Choices
AI assistants may leave out crucial warnings or details, subtly influencing your choices about health, money, and safety—often without you realizing it.
Read analysis PARENTING EDUCATIONAI Tutors Now Talk to Kids—But Who’s Watching the Screens?
AI tutors that see and talk through kids’ screens are changing how children learn, but parents and teachers must weigh benefits against privacy and quality risks.
Read analysis