Every AI tool, independently evaluated.

SchoolAI is an AI platform purpose-built for K-12 classrooms. Teachers create "Spaces" — assignment-scoped chatbots grounded in a topic — that students join with a code. The platform layers K-12-specific safeguards on top of frontier LLMs: live teacher dashboards showing every student conversation in real time, automatic flagging of concerning messages (self-harm, bullying, sensitive disclosures), teacher-controlled scope so the bot stays on task, and an admin console with a signed DPA. Because supervision, scope-control, and flagging are core to the product rather than optional add-ons, SchoolAI is a strong KidSafe-tier option for supervised classroom use across grades K-12.

K-8 adaptive math and reading platform with narrowly scoped AI-driven personalization, strong K-12 privacy posture, and district-grade admin controls.

Osmo is an interactive learning platform that pairs a physical reflector and hands-on manipulatives with iPad/iPhone apps, using on-device computer vision (a narrow, purpose-built form of AI) to recognize letters, numbers, drawings, and tangram pieces. It is one of the safest AI-adjacent products for young children: there is no chatbot, no generative text or image output, no open internet, and no social features. The main considerations are cost, screen time, and Tangible Play's corporate ownership by BYJU'S — parents and schools should confirm the current privacy posture with the vendor before large deployments.

Khanmigo is Khan Academy's nonprofit, OpenAI-powered AI tutor designed exclusively for K-12 education. It employs a Socratic coaching approach—guiding students to answers rather than delivering them—and layers multiple safety controls including automated content moderation, parent notifications on flagged conversations, daily session limits, and no advertising. As of mid-2026, Khan Academy is redesigning Khanmigo to embed AI assistance directly inside lessons.

Machine Learning for Kids is a free, teacher-led platform that lets children build simple machine-learning models (image, text, sound, and number classifiers) and use them in Scratch or Python projects. Created by Dale Lane and supported by IBM, it is one of the safest AI tools for children currently available: a teacher-managed account model eliminates the need for children to hand over personal information, there is no chatbot, no image generation, no advertising, and no commercial upsell. It is widely used in UK primary and secondary schools, CoderDojo, and Code Club, and is a strong fit for AI literacy under the EU AI Act's emphasis on transparent, educational AI experiences.

Canva Magic Studio is Canva's bundle of generative-AI features — Magic Write (text), Magic Design, Magic Media (image and video generation), Magic Edit, Magic Switch, and Magic Grab — layered into the Canva editor. Consumer Canva is 13+; Canva for Education is free for K-12 teachers and students with district provisioning, a signed DPA, and age-appropriate gating that turns off higher-risk AI features (like Magic Media image generation) for under-13 accounts by default. Strong safety posture for a creative AI product: content-safety filters on generation, watermarking on AI images, and teacher-controlled classroom settings. Tween Approved (blue) at score 84; safe for supervised classroom use in grades 4-12 through Canva for Education.

Google's source-grounded AI research assistant. NotebookLM answers only from documents the user uploads (PDFs, Google Docs, slides, web pages, YouTube transcripts, audio) and cites the exact passage behind every claim, making it one of the safer generative AI tools for structured study and research when used by teens under appropriate account terms.

LittleLit AI is a homeschool platform that states it is WASC-accredited — all-in-one K–12 homeschool curriculum built around age-appropriate AI tutors and auto-generated learning plans. It is purpose-built for children rather than adapted from a general-purpose chatbot, covering 16 core and enrichment subjects with structured lessons, projects, and progress reports for homeschools, microschools, and special-education programs.

Grammarly is a widely used AI writing assistant that provides grammar, spelling, clarity, and tone suggestions, plus generative writing features (rewrites, drafts, brainstorming) via GrammarlyGO. The consumer product is 13+ and Grammarly for Education offers district deployment with a signed DPA, SSO, and admin controls. Suggestion-based features are low-risk for K-12; generative features raise standard academic-integrity considerations and should be framed as revision, not authorship. Tween Approved when deployed through Grammarly for Education; safe for supervised classroom use in grades 6-12.

Gemini for Education is Google's AI assistant provisioned through the Google Workspace for Education tenant, with a signed DPA, no-training-on-student-data commitments, and admin controls that gate features per organizational unit. Consumer Gemini (gemini.google.com without an Edu account) is 18+ and does not carry these protections. When deployed through the Edu tenant with the recommended admin configuration, Gemini for Education is a strong option comparable to Microsoft Copilot for Education. A January 2026 Common Sense Media assessment flagged weak age-appropriate defaults and moderation gaps in the consumer/default configuration; those concerns are mitigated — not eliminated — by the admin controls documented below. Tween Approved (blue) at score 80; safe for supervised classroom use in grades 6-12 through Workspace for Education.

A math-problem-solving app now owned by Google. Point a camera at a handwritten or printed math problem and Photomath returns step-by-step solutions, explanations, and animated instructions covering elementary through college-level mathematics.

Google's consumer generative AI assistant with text, image, code, voice, and Deep Research capabilities. Available at gemini.google.com and in the Gemini app, with an optional AI Pro tier. Separate from Gemini for Education, which runs inside a Workspace tenant with different data protections.

Microsoft Copilot is a general-purpose AI assistant built on OpenAI models and integrated across Bing, Windows, Edge, and Microsoft 365. The consumer version is 18+ and comparable to ChatGPT for K-12 risk. The school-facing version — Copilot Chat / Microsoft 365 Copilot Chat provisioned through a Microsoft 365 Education (Entra ID) tenant — extends enterprise data protection, keeps prompts and responses out of foundation-model training, and is offered under Microsoft's OST / DPA with FERPA and student-data commitments. Teen Approved when deployed through the Edu tenant with admin controls; not appropriate for direct student use on the consumer product.

MagicSchool AI is a purpose-built K-12 AI platform offering 80+ teacher tools and 50+ MagicStudent tools with genuine privacy infrastructure: SOC 2 Type II, iKeepSafe FERPA/COPPA certification, and a 95% Common Sense Privacy rating earned March 2026. Student-mode requires school or teacher provisioning, keeps moderation always on, and explicitly prohibits LLM training on student data. A June 2026 Broward County School Board debate is a real-world caution flag that governance processes must accompany the technology.

CENTURY Tech is a UK-based AI-powered adaptive learning platform founded in 2013 and deployed across 1,000+ schools and colleges, primarily in England and Wales. It operates in a school-as-controller/Century-as-processor model that strongly limits independent data use, and its May 2024 and May 2026 privacy policy updates reflect active compliance work. SafeGradeAI's prior 'not school-safe' classification is overturned by current evidence for UK school use; meaningful caveats remain around the ICO June 2026 sector audit and the absence of US COPPA certification.

Quizlet's AI study features (Q-Chat, Magic Notes, practice-test generation, Explain-my-answer) sit on top of the long-running Quizlet flashcards platform. Widely used across US middle schools, high schools, and colleges, with free and Quizlet Plus tiers.

Anthropic's Claude is a leading frontier LLM chatbot built by a safety-focused AI lab, with strong published policies, an industry-leading Constitutional AI approach to moderation, and unusually transparent model documentation. However, Anthropic's own consumer terms require users to be 18+, the product is not designed for children, and there is no COPPA-compliant K-12 product for under-13 students. Claude for Education is aimed at higher education, not K-12. For SafeGradeAI purposes it is not recommended for direct student use in K-12; it is a reasonable adult tool for educators when used under district policy.

Prodigy Education is a widely deployed K-8 adaptive math and English platform used by tens of millions of students. It has a FERPA-facing school offering, published COPPA operator notices, and closed, moderated in-game content — the child-facing surface itself is safe. What holds Prodigy back is not privacy or moderation but design ethics: the free product is engineered around a Premium Membership upsell aimed at children, and Prodigy has been the subject of FTC scrutiny (2021), a UK ASA ruling (2022), and repeated researcher/parent criticism for pressuring minors to ask parents to pay. It is broadly school-safe as a supervised classroom tool; at home, parents should expect and manage the upsell pressure.

Brainly is a large peer-plus-AI homework help platform. The AI Learning Companion generates step-by-step explanations, summaries, and practice questions on top of Brainly's community answer corpus. Free tier is ad-supported; Brainly Plus/Tutor tiers unlock full AI and expert answers.

Duolingo Max is the premium AI tier of Duolingo, layering GPT-4 Turbo-powered Roleplay and Video Call features on top of the gamified lesson loop. The AI scenarios are tightly scoped to real-world language tasks, which meaningfully limits off-topic outputs. However, audio and transcript data from AI conversations are retained and used to train Duolingo's own models, the platform has declined to sign student data privacy agreements required by many US districts, and its heavy gamification mechanics raise engagement-design concerns for minors.

Poe (by Quora) is an aggregator that gives users access to many underlying models (GPT-4, Claude, Gemini, Llama, Mistral, image and video models) plus a marketplace of user-created custom bots. Free tier and paid subscription tiers unlock higher-capability models.

A generative AI music platform that produces full songs — vocals, instrumentation, lyrics — from a text prompt. Free tier with daily credits, plus Pro and Premier subscriptions. Rapidly popular with teens for social sharing.

Perplexity is a consumer AI answer engine that combines an LLM with real-time web search and inline citations. Source-grounded answers make it stronger than most chatbots for information literacy, but the consumer product is built for adults: there is no age verification, no parental controls, and no COPPA-aligned data handling. Perplexity for Enterprise offers a DPA, SOC 2 Type II, and data-training opt-out, but the free/Pro consumer tiers are not appropriate for children and are not designed for K-12 district deployment.

ChatGPT Free is OpenAI's zero-cost consumer chatbot tier, powered by rate-limited GPT-5.5 Instant as of mid-2026. It is a capable general-purpose AI assistant designed primarily for adults, with teen-oriented safety features bolted on in late 2025. However, the free tier's default opt-in to conversation training data, lack of institutional controls, easily circumvented parental controls, and absence of a verified age gate make it unsuitable for children under 13 and requiring careful supervision for teens.

Meta's Llama-powered assistant, embedded across Instagram, Messenger, WhatsApp, Facebook, and meta.ai, plus AI Studio persona chatbots. Available to any user 13+ on Meta apps; conversations sit alongside a social graph and advertising ecosystem.

ChatGPT (consumer Free and Plus tiers) is OpenAI's flagship general-purpose AI chatbot. While OpenAI made meaningful safety improvements since mid-2025 — parental controls (Sept 2025), safety-routing, a 170-expert mental-health update (Oct 2025), and Trusted Contact (May 2026) — those changes came reactively, following wrongful-death lawsuits involving minors. The consumer product still relies on self-reported age, trains on conversations by default, and was designed for adults; it is not recommended for children or unsupervised teens.

xAI's chatbot, positioned as a 'rebellious' assistant with real-time access to X posts. Available at grok.com, in the Grok mobile app, and inside X. Offers 'Fun Mode', an image generator with looser content policies than peers, and a 'spicy' mode in the Grok app that produces sexually suggestive content and near-nude imagery.

A GPT-powered chatbot pinned to the top of every Snapchat user's chat list, including 13–17 year-olds by default. My AI is embedded in a social/ephemeral messaging product optimized for engagement, and behaves as a persistent conversational companion for teens.

A Chinese generative AI assistant (chat.deepseek.com and DeepSeek app) built on DeepSeek's open-weight R1 and V3 models. Popular for reasoning and coding performance, but subject to Chinese data-protection law and content restrictions that make it a poor fit for K–12 use.

Squirrel AI (Yixue Education) is a Shanghai-headquartered K-12 adaptive tutoring platform with over 3,000 learning centers, more than 43 million cumulative students, and recognized by TIME100 2026. Its AI-driven personalization and peer-reviewed efficacy research are genuine strengths. However, the company's deep ties to the People's Republic of China — combined with extensive collection of facial biometrics, voice data, precise geolocation, and behavioral/psychological profiles of children — creates serious data-sovereignty and child-safety risks.

Manus AI (manus.im) by Butterfly Effect is a fully autonomous AI agent capable of browsing the web, writing and executing arbitrary code, and managing files — all without per-action user approval. It is restricted to adults 18+ by its own terms, carries documented critical security vulnerabilities including zero-click prompt-injection exploits (SilentBridge), and was founded by Chinese developers who relocated to Singapore, creating unresolved data-sovereignty concerns. It is wholly unsuitable for minors or educational environments.

Open-ended AI companion platform. Character.AI removed open-ended chat for under-18 users in late 2025 and now runs age assurance; SafeGradeAI's editorial recommendation remains 18+.