Government Employees to stop using AI Tools like Chatgpt, Deepseek
Asia News Agency
The government has issued a directive banning the use of AI tools and applications, such as ChatGPT and DeepSeek, on official government devices. The circular, dated 29th January 2025, aims to protect confidential government data from potential security threats.
This ban is part of a wider global apprehension regarding AI platforms handling sensitive data. Many AI models, including ChatGPT, process user inputs on external servers, raising concerns about data leakage or unauthorised access.
Similar AI restrictions have been imposed by governments and corporations worldwide. Several private companies and global organisations have already limited AI tool usage to prevent data exposure.
Reasons why the government could be taking this step
Below are some reasons why the government could be taking this step:
Risk of data leaks: AI models like ChatGPT and DeepSeek process user inputs on external servers, which means any sensitive government data entered into these tools could be stored, accessed, or even misused. Since government offices handle classified financial data, policy drafts, and internal communications, even unintentional sharing could pose risks.
Lack of control over AI models: Unlike traditional software used in government offices, AI tools are cloud-based and owned by private companies (such as OpenAI for ChatGPT). The government has no direct control over how these tools store or process information, increasing concerns about foreign access or cyber threats.
Compliance with data protection policies: India is working on stronger data privacy laws, including the Digital Personal Data Protection (DPDP) Act, 2023. Allowing AI tools on office devices without clear regulations could lead to violations of data protection policies, making government systems vulnerable.
India to get its first homegrown AI foundational model in 10 months
Meanwhile, India is set to launch its first homegrown AI foundational model within the next 10 months, according to IT Minister Ashwini Vaishnaw. This initiative aims to bolster India's position in the global AI landscape, which is currently dominated by the US and China.
Vaishnaw emphasized the government's dedication to AI development through the India AI Mission, which was approved last year with a budget of ₹10,000 crore. The mission seeks to make AI technology and computing power accessible to researchers, startups, and academic institutions, enabling them to contribute actively to the field.
Providing shared compute resources: A major focus of this initiative is the expansion of computing infrastructure. Vaishnaw revealed that India has already established a network of 18,000 high-end GPUs, with 10,000 available for AI development. These resources will be accessible to researchers, universities, and startups, aiming to lower financial barriers to AI innovation. "The first and most critical requirement for building AI models is compute power," said Vaishnaw. "Either people with deep pockets buy it, or we create a structure where the government ensures it is accessible to everyone."
By providing shared compute resources, the government intends to democratise AI development, similar to the Digital India programme. This system will enable Indian institutions to develop foundational AI models, which serve as the base for applications like generative AI, machine learning tools, and automation technologies.
With a completion timeline set at an "outer limit" of 10 months, India is advancing in a sector that is crucial for economic growth and technological self-reliance.
Involving the non - governmental sector in innovation
At a larger level, R A Mashelkar (president, Pune International Centre) Susan Thomas and Ajay Shah (co-founders at XKDR Forum) underscore the need for involving non-government institutions and organisations in science and research.
Their research suggests there are big gains in placing taxpayer resources into private firms and universities.
In France, defence research is funded by the government and happens in private defence firms. In the US, 80 per cent of NASA’s budget is contracted out to private firms and universities. NASA does not make spacecraft: Private vendors do. The recent accomplishment of Deepseek in China shows how innovative energy comes about in a country: The (private) team that created DeepSeek cut their teeth in computer science doing algorithmic trading.
Indian “will be higher levels of societal gains when knowledge resides in private organisations, and feeds into society.”
Contracting-out in the Indian context: The three authors sketch details of how to do this contracting-out in the Indian context.
“There is a natural connection between risk in research, and contracting out to private persons. Government can contract out the same research problem to multiple implementers who take different pathways. Some pathways would fare poorly, and the flow of money into them would be stopped. Such unfolding of risk is harder when laboratories and bureaucracies are built in government organisations.
“When a private firm is working in an area (for example, an automobile component firm that’s challenged to produce ball bearings of a superior spec), it will try to do the research well, because it also has a direct interest in the knowledge sought to be produced. Such a firm would bring knowledge from its normal operations into the contracted research; it would care deeply about the work and execute it well, and then the knowledge produced through the publicly funded research would spillover into its economic success.”
Many in Indian science policy have, of course, been thinking on similar lines. An important new organisation, Anusandhan National Research Foundation (ANRF), will put out grants at about Rs 2,800 crore a year through new ways. “It is likely to break new ground in getting public money to private organisations that will fund early stage research.”
The recent budget announced the allocation of Rs 20,000 crore ‘to implement private sector driven research, development and innovation initiative.’ This is an important milestone in the history of Indian science policy, note the three authors.
For example, ISRO will buy launch vehicles made by private persons. Taxpayer money will go to private firms who will do cutting-edge engineering, and the knowledge will feed into civilian applications and global competitiveness.
In recent years, the Ministry of Electronics and Information Technology (MEITY) has worked on using public funding to get more GPUs into private hands, to deepen AI knowledge in India. MEITY has procured 18,693 GPUs, which will be in operation at multiple private IT infrastructure firms. MEITY will give the use of these GPUs to researchers in Indian private organisations at the price of $1 per hour. “This is the philosophy of buy, not make,” write the three authors.
These developments, “will shape up as an important turning point for Indian science policy. The Indian state is rising out of the concept of using public money to hire researchers who are civil servants, to the concept of delivering public money into private universities and firms where cutting-edge knowledge is produced. This is greater bang for the taxpayer’s buck, as opposed to vertical government science organisations.”
