Major AI companies and education institutions are clashing over how to safely integrate artificial intelligence into classrooms, even as safety researchers raise alarms about existing AI tools harming young users. The tension reflects a broader challenge facing the education sector: AI adoption is accelerating rapidly, but guardrails and best practices are still being debated.

What Safety Concerns Are Experts Raising About AI in Schools?

Common Sense Media, a nonprofit focused on children's media and technology, conducted a comprehensive test of Google's AI Overview feature and found troubling results. After running 2,600 queries, researchers discovered that the AI search tool missed crisis signals, completed all 180 homework tasks tested without resistance, and families and schools cannot disable the features on accounts belonging to minors. The organization is now calling for default restrictions on accounts belonging to minors.

Beyond search tools, a new policy framework called the Flourishing Generation whitepaper proposes treating AI companions and human-like chatbots designed to encourage emotional dependence as adult products by default. The framework also recommends pre-market and post-market safety testing for any generative AI (artificial intelligence that creates text, images, or other content) designed for children.

These concerns arrive as educators grapple with a separate problem: AI-enabled cheating. As some AI tools fuel the proliferation of digital cheating, there's a reckoning happening among educators about how they evaluate and assess learning through traditional assignments, papers, and homework.

How Are Major Tech Companies Responding to Safety Concerns?

OpenAI has expanded ChatGPT's teen controls with a new Study Mode and safety notifications. Parents can now make Study Mode the default for linked teen accounts, and OpenAI is extending notifications to cases where accounts are deactivated for violent threats or acts of violence online, alongside more frequent break reminders.

OpenAI also launched a broader Public Policy Agenda setting out six priorities spanning frontier model safety, protections for minors, education, workforce change, content provenance, and infrastructure. Its proposals include teacher training, age assurance, independent audits, and regional AI skills hubs.

Meanwhile, major universities are partnering with AI companies to build education-specific tools. Western Governors University and Anthropic announced plans to develop AI-native learning and credentialing systems for identifying workforce skills, personalizing learning, and verifying capabilities through digital credentials. Anthropic will provide Claude for Enterprise, model credits, engineering support, and platform resources. Similarly, University of Phoenix and OpenAI are collaborating to explore ChatGPT Edu, AI-enabled teaching, student support, career services, and research into learning and workforce outcomes.

Why Are Schools Struggling to Implement AI Responsibly?

A central problem is that educational institutions in the United States have been underfunded for decades, which creates systemic strain. When systems face strain, administrators often look for ways to cut costs and become more efficient, even though education fundamentally requires human judgment and relationship-building.

"Whenever you're thinking about technology in education, you should really think about it like a three-legged stool. You have the actual technology, you have the person using it, and then you have the place where they're using it," said Stephen Aguilar.

Stephen Aguilar, Associate Professor at the USC Rossier School of Education

Aguilar explained that in an ideal world, the technology works, the person knows how to use it, and it's used in a place where it actually makes a difference. Much of the backlash against AI in schools stems from poor fit, where AI or EdTech companies enter educational spaces they know very little about.

Emily Weinstein, a researcher at the Center for Digital Thriving at Harvard, noted that some students are using up to seven different apps to manage everything from grades to sports team schedules. Kids have reported feeling constantly tethered to these tools to monitor the "endless stream of notifications about what they're supposed to know and do," which can add to the academic pressures they're already experiencing.

Steps Schools Can Take to Implement AI More Thoughtfully

• Establish Clear Policies: Four more states now require districts to adopt AI policies, signaling a shift toward structured governance. Schools should develop clear guidelines about which AI tools are permitted, how they'll be used, and what safeguards are in place.
• Prioritize Teacher Training and Support: OpenAI's policy agenda emphasizes teacher training as a core priority. Schools should ensure educators understand how to use AI tools effectively and ethically before deploying them in classrooms.
• Focus on Academic Integrity Over Detection: Rather than layering on detection tools to catch cheating, schools should create environments that encourage academic integrity and help students understand the purpose of learning. Some leading EdTech platforms like Instructure and Inspera have rolled out tools to help instructors understand the culture around cheating and shift focus from catching students to creating integrity-focused environments.
• Involve Students in the Conversation: Listening to student perspectives could fundamentally change how schools understand what it looks and feels like to be a student in an AI-rich context. When you sit down with a young person and examine their experience, you can see that their good intentions are not always enough in a context where market forces and design features are pulling them in different directions.

What Does the Broader EdTech Landscape Look Like Right Now?

Despite rapid AI adoption, the EdTech market remains dominated by platforms that predate the AI boom. TIME and Statista partnered on a data-based research project evaluating thousands of companies in the space on strong financial performance and industry impact, resulting in a list of the 500 top-performing companies in education technology. The top-ranked company is Duolingo, which started as a language learning app and has expanded into math, chess, and music. Duolingo now has over 60 million daily active users and generated more than $1 billion in bookings last year.

Duolingo's success illustrates a key principle: engagement matters. The platform stands out with engaging designs that motivate users to return and practice every day, micro-lessons that use implicit learning rather than drilling grammar rules, and a free tier that brings in new users. Around half of Duolingo's users are under 25, and use the app as an educational supplement.

However, in the United States, where there appears to be a "learning recession" as standardized scores for math and reading drop, there are rising concerns about the use of technology in classrooms and too much screen time for children. This tension between AI's potential and its risks will likely define education policy for the next several years.