India Was the World's First GPT-4 Test Market, Without Knowing It: Report

On February 19, 2026, roughly three and a half years after the concealed deployment, OpenAI CEO Sam Altman stood at the India AI Impact Summit in New Delhi and called for an international body to govern AI responsibly.

Advertisement
Read Time: 3 mins
OpenAI and Microsoft had set up a joint safety board whose job was to review new AI models before public release.

In 2022, Microsoft quietly tested GPT-4 on Indian users in 2022 through its Bing search engine, making India the first country in the world where real humans were exposed to the AI model. The users had no idea. Neither, it turns out, did OpenAI's board.

That is the central finding of an April 7 investigation in The New Yorker by journalists Ronan Farrow and Andrew Marantz, based on over 100 interviews and hundreds of pages of internal memos from former OpenAI chief scientist Ilya Sutskever and Anthropic co-founder Dario Amodei.

Advertisement

ALSO READ: Coforge Shares In Focus After Approval Of Key AI Acquisition

How It Happened?

OpenAI and Microsoft had set up a joint safety board whose job was to review new AI models before public release. Reportedly, Microsoft had not sought clearance from that board for the India test. OpenAI had in fact warned Microsoft not to go ahead. Microsoft went ahead anyway.

The breach was not just kept from the board, it was kept from OpenAI's leadership too. According to the investigation, as board member Tasha McCauley was wrapping up a six-hour OpenAI board meeting, an employee pulled her aside to quietly tell her what had happened. CEO Sam Altman had said nothing about it during the entire meeting. Jacob Hilton, an OpenAI researcher at the time, told the journalists: it was "kind of completely ignored."

When the India deployment was first reported elsewhere, Microsoft flatly denied it. Spokesman Frank Shaw told The New York Times that GPT-4 had not been used in the India tests. But, after The New Yorker piece was published, Microsoft reversed course and confirmed that GPT-4 had indeed been tested on Bing in India.

Advertisement

ALSO READ: AI Threat For Jobs: TCS Top Management Doesn't Believe So, Says ' Hiring More Advisors'

The timing carries a certain irony. On Feb. 19, 2026, roughly three and a half years after the concealed deployment, OpenAI CEO Sam Altman stood before ministers and tech executives at the India AI Impact Summit in New Delhi and called for an international body to govern AI responsibly. India today has 180 million monthly ChatGPT users and became the world's largest market for generative AI app downloads in 2025, with installs rising 207 per cent year-on-year.

Essential Business Intelligence, Continuous LIVE TV, Sharp Market Insights, Practical Personal Finance Advice and Latest Stories — On NDTV Profit.

Loading...