India’s Public School Teachers are Adopting AI—and Outrunning Rules

Before her class, Amrita (name changed), a teacher at a government school in Karnataka, found herself struggling to prepare. The topic was complex and required breaking down concepts that were outside the structure of the syllabus. She picked up her phone a day before the class and asked an AI tool to help her explain the concept. 

She wasn't trained to do this. Nobody at her school gave her any guidelines on which tool to use, how to use it, and what prompts to give.

"AI can help me explain difficult topics in a simple way, create examples, and prepare activities according to students' learning levels," Amrita tells AIM, describing what has quietly become part of her routine.

She is among a vast swathe of teachers and educationalists who are quietly but permanently imbibing AI tools into their professions.

A new survey of 797 government school teachers across Uttar Pradesh, Karnataka, Jharkhand, and Maharashtra, released by JanAI, an initiative of Head Held High Foundation to set up community AI centres in rural India, found that 82% had tried an AI tool at least once. Moreover, 72% felt positive about AI in education.

Yet the schools where many of these teachers work still have limited access to basic computing infrastructure. Only 44% of government schools had a functional computer in 2023–24, compared with 71% of private schools, according to UDISE+ data cited in the report.

The finding comes as institutions such as CBSE and NCERT prepare to introduce an AI and Computational Thinking curriculum for Classes 3 through 12, phased in from the 2026–27 academic session. 

The Teacher Got There First

Despite the infrastructure gap, the report noted that 51% of teachers at schools with no computer lab were using AI in some capacity. At schools with a working, regularly used lab, that figure rose to 72%. Around 24% of teachers who had never used AI started using it professionally.

Separately, BaSE 2025, the Central Square Foundation's (CSF) national survey of 2,500 government, government-aided and private-school teachers, found that 83% of teachers who were aware of or using edtech had heard of AI. More than half, 51%, were already using generative AI tools for teaching or school-related work.

Gouri Gupta, Senior Project Director, Edtech and AI at Central Square Foundation, says adoption is happening "faster than most systems have caught up with."

The pattern also cuts across school types. GenAI use stood at 68% among government-aided teachers and 51% among government-school teachers, compared with 40% among private-school teachers.

Half of the GenAI-using teachers in the CSF survey discovered the tools through their own school or fellow teachers rather than through a formal channel.

High Trust, High Concern

The findings point to the trend of teachers increasingly relying on AI tools as they believe it makes them personally more effective. However, they also worry about what unsupervised use could do to their students.

Per the JanAI survey, 86% of teachers who reported a personal benefit from AI expressed concern about student misuse, compared with 64% of those who did not see the same benefit.

"My biggest concern is that students may become overly dependent on AI and stop developing their own thinking and creativity," one teacher in Uttar Pradesh expresses. 

Teachers are finding practical uses for AI, such as simplifying difficult concepts, generating examples and activities, and preparing worksheets and assessments.

But using AI for those tasks requires understanding what the systems are actually doing. That is where the gap becomes more significant.

Adoption + Awareness

CSF's BaSE 2025 survey found that only 46% of teachers who knew about GenAI understood how it works. This points to the need for AI literacy. 

The JanAI findings also showed that teachers who had received formal training were not more likely to verify AI-generated answers than those who had not.

But the problem may run deeper, including what the teachers are actually getting trained on.

According to CSF, 51% of edtech teachers had received some technology-related training during the previous year. But only 9% of that training was specifically about AI tools. The largest portions covered basic digital literacy, safe technology use, and specific tools or devices.

"Most 'trained' teachers were perhaps trained on broader digital literacy and not necessarily on AI itself," Gupta comments.

However, even after training, adoption isn’t uniform. Most stop using AI tools after one or two sessions.

Per JanAI, positive sentiment stood at 61% among never-users, fell to 49% among teachers who remained stuck after one attempt, then climbed to 79% among occasional users and 91% among regular users. The pattern suggests that the first experience with AI can determine whether experimentation becomes a habit.

The same problem appears in rural areas. Rural teachers tried AI at a higher rate than urban teachers, 29% compared with 15%, but were less likely to become regular users, at 19% compared with 31%.

What Teachers are Inputting

A bigger risk sits beneath the adoption numbers.

More than half of teachers who use AI tools, 51%, said they had entered a student's name, marks or photograph into an AI tool at some point.

The exposure was higher among teachers with leadership or administrative responsibilities. Around 64% had entered student data, compared with 45% of classroom teachers without those responsibilities.

"I worry about data privacy and the security of students' personal information when using AI tools," a Karnataka teacher remarks anonymously.

The problem is exacerbated amid the lack of school guidelines on AI use and student privacy. 

CBSE, with support from Intel, has curated guidelines on AI integration in school curriculum, building a framework on effective pedagogy, assessment approaches, and interdisciplinary integration. While National Education Policy 2020 has introduced AI for students, schools lack a unified national framework or protocol for teacher usage, academic integrity, and data privacy. 

What Actually Moves the Needle

The JanAI survey found that institutional support made a substantial difference.

Researchers scored schools on five forms of support: school rules, state policy, whether teachers had read available guidance, whether training had been offered, and whether someone was available to help. At schools scoring zero on those measures, 49% of teachers had a positive view of AI. At schools scoring five, that figure rose to 97%.

Regular AI use also increased from 13% to 59%.

One intervention stood out. Teachers who had discussed responsible AI use with their students were four times more likely to say they were very confident distinguishing genuine student learning from AI-assisted copying, at 39% compared with 10%.

School curricula are now catching up to habits teachers have already developed on their own.

For organisations like CSF, this gives an opportunity to expand AI literacy and integrate it into the education system.

Its AI Samarth initiative focuses on helping teachers and students understand how AI works, question and verify its outputs, recognise bias and misinformation, protect privacy, and know when human judgement should take precedence. The initiative has reached 2.6 million learners across more than 10 states, according to CSF.

Meanwhile, JanAI plans to build a community of 50,000 rural government school teachers as local AI champions over the next 12–18 months. The programme will work state by state, using the survey findings to shape interventions around responsible use, guidance, and support.

The question is whether schools can build the literacy and support systems quickly enough to guide teacher habits before AI becomes a formal part of every classroom.