CBSE AI Curriculum: What Classes 3–8 Learn and Free Tools

The CBSE AI curriculum that started this academic year is smaller and better designed than the headlines suggest. For Classes 3 to 5 it is not a new subject at all: it is computational thinking folded into Maths and "The World Around Us", taught by the same teachers from worksheets, and assessed inside those subjects. AI itself arrives in Classes 6 to 8, for 20 hours a year, with the computer teacher. Here is what CBSE's own framework says, what it costs (nothing — the handbooks are free downloads), and which free tools match it.

What CBSE actually announced, and when

The sequence matters because most write-ups blur it.

On 11 October 2025, School Education Secretary Sanjay Kumar said AI would be introduced for all students from Class 3 in 2026-27, with CBSE developing the framework, per PTI. The same report noted that over 18,000 CBSE schools already offered AI as a skill subject from Class 6 through a 15-hour module, optional in Classes 9–12.

CBSE then published the Computational Thinking and Artificial Intelligence Classes 3–8 Curriculum, with a student and a teacher handbook for every class from 3 to 8 — first edition March 2026 — on its CT&AI page, alongside a circular and a recorded webinar.

On 6 August 2026, in a written Rajya Sabha reply reported by NE India Broadcast, Minister of State Jayant Chaudhary confirmed the Classes III–VIII rollout from 2026-27, said the curriculum for grades 9–12 "has been finalized", and noted that AI and robotics labs are the responsibility of state governments.

The CBSE AI curriculum, stage by stage

The framework's own numbers:

	Classes 3–5 (Preparatory)	Classes 6–8 (Middle)
Hours a year	50	100
What it is	Computational thinking embedded in Maths and The World Around Us	Advanced CT, data handling, AI awareness, ethics
How the hours split	Within existing subjects	40 CT, 20 introductory AI, 40 interdisciplinary projects
Who teaches	Subject teachers, from worksheets and handbooks	Subject teachers for CT; the computer teacher for AI literacy
Materials	A resource book of CT questions following the textbook chapters	A resource book with complex CT, AI foundations and project details, plus teacher manuals
Assessed how	Inside Maths and TWAU	Projects, assignments, reflective journals, assigned tasks

The framework is explicit that "CT forms the basis of learning AI", so the early years are puzzles, visual reasoning and "basic problem-solving in everyday contexts" — not algorithms, not code, not chatbots. It also states that the design is "without dependence on any specific platform."

What "AI" means by Class 8

The learning outcomes for Class 8 are where the curriculum shows its hand, and they are more careful than the phrase "AI for eight-year-olds" implies. Learners should be able to:

• Describe the AI project cycle: "Define Problem, Collect Data, Test AI Tools, Reflect and Improve"
• Apply no-code tools to real-world problems and reflect on their usefulness
• Explain how AI uses data, "find and research sources of bias in datasets", and apply basic strategies for fairness
• Understand AI as "a specific type of algorithm that uses datasets, learning, and prediction"
• Describe AI ethics as "the values and guidelines that ensure AI is created and used responsibly"

Bias, privacy and accountability are in the outcomes for a 13-year-old. That is in line with UNESCO's 2024 AI competency framework for students, which organises what children should learn into four dimensions — a human-centred mindset, ethics of AI, AI techniques and applications, and AI system design — and which, UNESCO notes, only 15 countries had built into national curricula by 2022.

The 20 AI hours, class by class

The syllabus for the AI component is four five-hour units a year, and it builds in a straight line:

Class	The four units
6	What AI is and everyday examples, including the difference between AI and automation and the three kinds of learning; basic data types; simple pattern recognition; online safety, privacy and digital footprints
7	AI domains — classification, regression and clustering practised on a small dataset, plus computer vision, NLP and data science; AI in healthcare, education and transport; charts and data analysis; bias in AI and digital citizenship
8	The AI project lifecycle; hands-on work with "simple no-code AI tools (image classifiers, chatbots, data prediction apps)"; data and fairness; privacy, misinformation and responsible use

Two things stand out. Supervised, unsupervised and reinforcement learning are named in Class 6 — as concepts, not maths. And ethics is a unit in every year, not a closing lecture.

How children are assessed

Not by a new board exam. The framework says assessment "needs to move away from traditional summative assessment to continuous, formative, and competency-based assessment", and lists the tools: problem-solving tasks, project work, reflective journals, "discussions/debates on ethical dilemmas", and teacher observation. For Classes 3–5 it suggests written tests with CT puzzles and interactive group activities such as treasure hunts. The stated focus is applying knowledge and developing ethical reasoning "rather than assessing rote memorisation".

For Classes 6–8, the interdisciplinary projects — 40 of the 100 hours — are assessed by the computer teacher.

Free tools that match the syllabus

The framework "recommends harnessing free and open-source tools to ensure accessibility and adaptability for all institutions", and the Class 8 outcome about "no-code tools" points at a specific kind of software. These are the ones that fit each stage, all free to use.

Scratch — for the computational-thinking years. Block-based coding from the Scratch Foundation, developed at MIT in 2007, free for everyone, used by 200 million children. Decomposition, patterns and algorithms are what you learn by building a game in it.

Teachable Machine — the closest thing to the Class 8 project cycle in one free tool. Google's browser tool trains a model to recognise images, sounds or poses from your webcam, with no coding and no account; per Google's blog, "all the training happens in your browser, so everything stays in your computer." Collect data, test, notice it fails on the examples you didn't include, improve. That is the bias lesson, taught by experience.

Machine Learning for Kids — built by IBM developer Dale Lane as a side project for local schools, it lets children train models on text, numbers or images and use them inside Scratch, per its project page. It bridges the two tools above.

Code.org's AI courses — AI for Oceans, for grades 3–12, has students train a model to sort sea creatures from trash; AI Discoveries, for grades 6–8, is listed as free and covers "using, thinking about, and creating with AI."

Day of AI — a free, grade-by-grade PreK–12 AI literacy curriculum from MIT RAISE, launched 19 August 2026 and reporting over five million students reached, at dayofai.org. Useful for teachers building the 20 AI hours.

None of these is an assistant that answers homework. For which of those a child should use, and at what age, see AI for kids: what's safe at each age.

What parents can do at home

Three things follow from the framework.

Don't buy anything for Classes 3–5. The stage is worksheets inside existing subjects. A child who likes puzzles is already doing the curriculum.

For Classes 6–8, a browser is enough. Teachable Machine and Scratch run in one. The framework recommends free tools on purpose, and labs are the state's job, not yours.

Keep the assistants separate. Learning what AI is and using a chatbot are different activities, and Common Sense Media's guidance draws the line by age: no chatbots under 5, supervised use from 6 to 12, independent use at 13–17 only for schoolwork. A Class 6 child can train a Teachable Machine model without ever opening ChatGPT.

Teachers have their own page here: AI tools for teachers, and the older-student version is AI tools for students. If a Class 8 project needs real code, our AI coding tools for beginners guide and free AI coding tools cover the browser-based editors.

Classes 9–12: what is and isn't settled

The existing AI skill subject continues — it is the one 18,000-plus schools already run — and the minister's reply says the 9–12 curriculum is finalised. What has not been published in a form we could read is the detail: whether CT&AI becomes compulsory in Class 9 and when, and what happens to the current elective. Secondary sources contradict each other on this. Check CBSE's own circulars rather than a coaching site.

FAQ

Is AI compulsory in CBSE from Class 3?

Computational thinking is, from 2026-27, but as part of Maths and The World Around Us rather than a separate subject — 50 hours a year. AI concepts themselves start in Class 6, for 20 hours a year within a 100-hour CT and AI allocation.

Is there an exam for the CBSE AI curriculum?

Not a new one. The framework calls for continuous, formative, competency-based assessment — projects, reflective journals, group activities and written CT puzzles — and says the focus is applying knowledge rather than rote memorisation. In Classes 3–5 it is assessed within the core subjects.

What is computational thinking?

CBSE's definition: "a structured approach to problem-solving that breaks larger problems into smaller, logical pieces, building precise, step-by-step solutions that either a person or a machine can follow." It is taught through puzzles and visual reasoning before any technology appears.

Where are the CBSE CT&AI handbooks?

On CBSE's CT&AI page at cbseacademic.nic.in, with a student and a teacher handbook for each of Classes 3 to 8, the curriculum framework, a circular and a webinar recording. They are downloads, not purchases.

Does my child need a laptop or a paid app?

For Classes 3–5, no — the stage is worksheet-based. For Classes 6–8 a shared browser covers the free tools the framework recommends. School AI labs, where they exist, are funded by state governments, not parents.